Barrick Gold (GOLD) 6-K Technical Report on the Kibali Gold Mine, Democratic Republic of the Congo

Exhibit 99.2

Technical Report on the Loulo-Gounkoto

Gold Mine Complex, Mali

Report for NI43-101

Randgold Resources Limited

	3rdFloor Unity Chambers 28 Halkett Street
	St Helier Jersey OJE2
	18thSeptember 2018
	Effective Date: 31stDecember 2017

Qualified Persons:

Mr. Rodney B. Quick MSc, Pr. Sci.Nat

Mr. Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM

Mr. Richard Quarmby, BSc, Pr Eng & C Eng, MSAIChE, MIoMMM, MBA

Mr. Derek Holm, BSc, FSAIMM

Mr. Graham E. Trusler, MSc, Pr Eng, MSAIChE

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table of Contents

1	Executive Summary		1
	1.1	Property Description and Location	1
	1.2	Mineral Rights and Land Ownership	2
	1.3	History	3
	1.4	Geology and Mineralisation	4
	1.5	Exploration	6
	1.6	Mineral Resources	7
	1.7	Ore Reserves	9
	1.8	Mining Method	11
	1.9	Mineral Processing	14
	1.10	Project Infrastructure	16
	1.11	Market Studies	17
	1.12	Environmental, Permitting and Social Considerations	18
	1.13	Capital Costs	19
	1.14	Operating Costs	20
	1.15	Economic Analysis	20
	1.16	Interpretation and Conclusions	21
	1.17	Risks	24
	1.18	Recommendations	29
2	Introduction		30
	2.1	Introduction	30
	2.2	Effective Date	30
	2.3	Sources of Information	31
	2.4	List of Abbreviations	32
	2.5	Units	34
3	Reliance on Other Experts		35
4	Property Description and Location		36
	4.1	Project Location	36
	4.2	Mineral Rights and Land Ownership	38
	4.3	Surface Rights	40
	4.4	Ownership, Royalties and Lease Obligations	41
5	Accessibility, Climate, Local Resources, Infrastructure and Physiography		42
	5.1	Accessibility	42
	5.2	Climate	42
	5.3	Physiography	43

18^th September 2018 Page ii

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

	5.4	Local Resources	43
	5.5	Infrastructure	44
6	History		47
	6.1	Historical Exploration and Development	47
	6.2	Randgold Project Milestones and Development	48
	6.3	Historical Resource and Reserve Estimates	50
	6.4	Resource and Reserve Evolution	50
	6.5	Past Production	52
7	Geological Setting and Mineralisation		53
	7.1	Regional Geology	53
	7.2	Local Geology	54
	7.3	Permit Geology	54
	7.4	Loulo Permit Geology and Mineralisation	56
	7.5	Gounkoto Permit Geology and Mineralisation	64
	7.6	Discussion	68
8	Deposit Types		69
9	Exploration		70
	9.1	Exploration Concept	70
	9.2	Loulo	71
	9.3	Gounkoto	72
	9.4	Discussion	72
10	Drilling		73
	10.1	Drill Hole Summary	73
	10.2	Survey Grids	74
	10.3	Drill Planning and Site Preparation	74
	10.4	Downhole Surveying	75
	10.5	Collar Surveys	75
	10.6	Diamond Drilling	75
	10.7	Reverse Circulation Drilling	76
	10.8	Trenching	78
	10.9	Other Sampling Methods	78
	10.10	Drill Twinning Studies	78
	10.11	Discussion	79
11	Sample Preparation, Analyses and Security		80
	11.1	Sample Selection	80
	11.2	Sample Preparation	80
	11.3	Sample Analysis	84

18^th September 2018 Page iii

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

	11.4	Quality Assurance and Quality Control	86
	11.5	Sample Security	101
	11.6	Independent Audit	102
	11.7	Discussion	103
12	Data Verification		104
	12.1	Independent Audits	105
13	Mineral Processing and Metallurgical Testing		106
	13.1	Summary	106
	13.2	Gounkoto Testwork	106
	13.3	Sampling and Sample Representativity	109
	13.4	Loulo Metallurgical Testwork	109
	13.5	Recovery	110
	13.6	Deleterious Elements	113
14	Mineral Resource Estimates		116
	14.1	Loulo Summary	116
	14.2	Gounkoto Summary	118
	14.3	Resource Database	119
	14.4	Geological Modelling	120
	14.5	Topography	129
	14.6	Bulk Density	129
	14.7	Compositing	134
	14.8	Treatment of High Grades (Top Cutting)	138
	14.9	Variography	142
	14.10	Block Model Estimation and Quantitative Kriging Neighbourhood Analysis	152
	14.11	Block Models	158
	14.12	Resource Classification	169
	14.13	Block Model Depletion	174
	14.14	Block Model Validation	174
	14.15	ResourceCut-Off Grades	180
	14.16	Other Minor Satellites	182
	14.17	Mineral Resources Reporting	183
	14.18	Comparison to Previous Models	188
	14.19	Reconciliation	193
	14.20	External Audits	194
	14.21	Discussion	195
15	Ore Reserve Estimate		197
	15.1	Introduction	199

18^th September 2018 Page iv

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

	15.2	Loulo Open Pit Ore Reserve	200
	15.3	Gounkoto Open Pit Ore Reserve	202
	15.4	Loulo Underground Ore Reserve	202
	15.5	Gounkoto Underground Ore Reserve	203
	15.6	Reserves Estimation Process	203
	15.7	Geotechnical Parameters	204
	15.8	Parameters Affected by Groundwater Control	208
	15.9	Dilution and Mining Recovery	210
	15.10	Economic Parameters	214
	15.11	Pit Optimisations	219
	15.12	Mine Design	225
	15.13	Ore Reserve Comparisons	231
	15.14	External Audits	232
16	Mining Methods		234
	16.1	Loulo Open Pits	234
	16.2	Gounkoto Open Pit Operations	237
	16.3	Loulo Underground Mining Operations	242
	16.4	Gounkoto Underground Operations	252
	16.5	Loulo-Gounkoto Life of Mine Plan	263
17	Recovery Methods		266
	17.1	Processing Plant	266
	17.2	Production History	271
	17.3	Capital Projects and Plant Upgrades	273
	17.4	Metallurgical Recovery	273
	17.5	Processing Costs	273
18	Project Infrastructure		275
	18.1	Mine Roads	276
	18.2	Supply Chain	276
	18.3	Surface Water Management	277
	18.4	Water Supply	278
	18.5	Tailings Facilities	278
	18.6	Power Supply	280
	18.7	Site Infrastructure	282
	18.8	Communication and Information Technology	283
	18.9	Security	284
19	Market Studies and Contracts		285
	19.1	Revenue, Tax and Royalty	285

18^th September 2018 Page v

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

	19.2	Markets	285
	19.3	Contracts	285
20	Environmental Studies, Permitting, And Social or Community Impact		287
	20.1	Environmental Considerations	287
	20.2	Social Considerations	290
21	Capital and Operating Costs		293
	21.1	Capital Costs	293
	21.2	Operating Costs	296
22	Economic Analysis		298
23	Adjacent Properties		299
	23.1	Tabakoto Gold Mine	299
	23.2	Kofi Exploration Project	299
24	Other Relevant Data and Information		300
	24.1	Country Risk	300
25	Interpretation and Conclusions		302
	25.1	Geology and Mineral Resources	302
	25.2	Mining and Ore Reserves	303
	25.3	Processing	304
	25.4	Environment and Social	304
	25.5	Ownership	305
	25.6	Infrastructure	305
	25.7	Risks	305
26	Recommendations		310
27	References		311
28	Date and Signature Page		314
29	Certificate of Qualified Persons		315
	29.1	Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM	315
	29.2	Rodney B. Quick MSc, Pr. Sci.Nat	316
	29.3	Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA	317
	29.4	Derek Holm, BSc, FSAIMM	318
	29.5	Graham E. Trusler, MSc Pr Eng, MIChE, MSAIChE	319
30	Appendix		320
	30.1	Appendix 1 – JORC 2012 Edition – Table 1	320

18^th September 2018 Page vi

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

List of Tables

Table1-1 Mineral Resource Statement for the Loulo Gold Mine as of 31stDecember 2017	8
Table1-2 Mineral Resource Statement for the Gounkoto Gold Mine as of 31stDecember 2017	9
Table1-3 Ore Reserve Statement for the Loulo as of 31stDecember 2017	10
Table1-4 Ore Reserve Statement for Gounkoto as of 31stDecember 2017	11
Table1-5 Actual Process and Plant Engineering Operating Costs for 2016 and 2017	16
Table1-6 LOM Capital Expenditure	20
Table1-7 LOM Operating Costs for Loulo and Gounkoto	20
Table1-8 Loulo Risk Rating	26
Table1-9 Gounkoto Risk Rating	28
Table4-1 Loulo and Gounkoto Permit Coordinates	39
Table5-1 Monthly Records of Precipitation and Potential Evaporation	43
Table6-1 Past Production Records for Loulo-Gounkoto	52
Table10-1 Loulo Drilling Summary	73
Table10-2 Gounkoto Drilling Summary	74
Table11-1 Summary of Certified Reference Materials Used at Yalea and Gara	84
Table11-2 Summary of Certified Reference Materials Used at Yalea and Gara	84
Table11-3 Summary of Certified Reference Materials Used at Yalea and Gara	88
Table11-4 Bivariate Statistics of Gara and Yalea CRM’s Assayed by SGS Loulo During the Period	88
Table11-5 Summary of Gounkoto Certified Reference Materials Used During the Period	91
Table11-6 Bivariate Statistics for Gounkoto CRMs Assayed by SGS Loulo	91
Table11-7 Loulo Blank Results Returned During Reporting Period	93
Table11-8 Gounkoto Blank Results Returned During Reporting Period	93
Table11-9 Bivariate Statistics for Log Scatter of Gara and Yalea Course Crush Rejects Duplicate Assayed by SGS Loulo During the Period	94
Table11-10 Bivariate Statistics for SGS Loulo Correlational Plot of Gounkoto Field Duplicates	97
Table11-11 Loulo Umpire Sample Summary	98
Table11-12 Bivariate Statistics for Loulo Umpire Assays	98
Table11-13 Loulo Umpire Sample Summary	101
Table13-1 Metallurgical Optimum Recoveries by Domain 2016	107
Table13-2 Gounkoto Testwork	108
Table13-3 Summary of Testwork Reports Prior to First Feasibility Study	110
Table13-4 Loulo Processing Plant Overall Gold Recovery in 2016 and 2017 by Month	112
Table14-1 Loulo Gold Mine Mineral Resource Statement as of 31stDecember 2017	116
Table14-2 Summary of Loulo Deposit and Model Date	117

18^th September 2018 Page vii

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-3 Gounkoto Gold Mine Mineral Resource Statement as of 31stDecember 2017	118
Table14-4 Summary of Gounkoto Deposit and Model Date	119
Table14-5 Loulo Mineral Resource Dataset	120
Table14-6 Gounkoto Mineral Resource Dataset	120
Table14-7 Yalea Density Data	130
Table14-8 Gara and Gara West Density Data	130
Table14-9 Loulo 3 Density Data	130
Table14-10 Baboto Density Data	131
Table14-11 Densities Applied Gounkoto for 2017 Resources	133
Table14-12 Densities Applied Faraba for 2017 Resources	133
Table14-13 Yalea Total Model 2 m Composite Dataset	134
Table14-14 Gara Total Model 2 m Composite Dataset	135
Table14-15 Loulo 3 Total Model 1 m Composite Dataset	135
Table14-16 Baboto South and Centre 1 m Composite Dataset	136
Table14-17 Gara West Total Model 1 m Composite Dataset	136
Table14-18 Gounkoto 2 m Compositing Data	137
Table14-19 Faraba Composite Sample Analysis	137
Table14-20 Yalea Top Cutting Values Applied to Composites	138
Table14-21 Gara Top Cutting Values Applied to Composites	139
Table14-22 Loulo Top Cutting Values Applied to Composites	140
Table14-23 Baboto Top Cutting Values Applied to Composites	140
Table14-24 Gara West Top Cutting Values Applied to Composites	140
Table14-25 Gounkoto Top Cutting	141
Table14-26 Top Cutting Values Applied to Composites	142
Table14-27 QKNA Parameters from Yalea	154
Table14-28 QKNA Parameters from Gara	155
Table14-29 QKNA Parameters from Loulo 3	155
Table14-30 QKNA Parameters from Baboto	155
Table14-31 QKNA Parameters from Gara West	156
Table14-32 QKNA & Kriging Parameters for Gounkoto	157
Table14-33 Summary of QKNA for Faraba Domains	158
Table14-34 Yalea Block Model Variables	158
Table14-35 Yalea Block Model Extents	161
Table14-36 Yalea Estimation Domain Orientation	161
Table14-37 Gara Block Model Extents	163
Table14-38 Gara Estimation Domain Orientation	163
Table14-39 Loulo 3 Block Model Extents	164
Table14-40 Loulo 3 Search Ellipse Orientations and HG Constraining	164

18^th September 2018 Page viii

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-41 Baboto South and Centre Block Model Extents	165
Table14-42 Baboto South and Centre Estimation Domains	166
Table14-43 Gara West Block Model Extents	166
Table14-44 Gara West Sub Domain Ellipse Orientations	167
Table14-45 Gounkoto Block Model Extents	168
Table14-46 Summary of the Gounkoto Orientations and HG Constraint	168
Table14-47 Faraba Block Model Extents	169
Table14-48 Modelled Semi-Variogram Parameters for Faraba	169
Table14-49 Loulo Classification Criteria	170
Table14-50 Gounkoto Classification Criteria	173
Table14-51 Gara and Yalea Volume Reconciliation	175
Table14-52 Gounkoto Volume Reconciliation	176
Table14-53 Gounkoto Open Pit Data Comparison	177
Table14-54 2017 Loulo Open PitCut-Off Grade Calculations	180
Table14-55 2017 Loulo UndergroundCut-Off Grade Calculations	181
Table14-56 Gounkoto Open Pit 2017Cut-Off Grade Calculations	181
Table14-57 Gounkoto Underground 2017Cut-Off Grade Calculations	182
Table14-58 Loulo Total Declared Resources as of 31stDecember 2017	186
Table14-59 Gounkoto Total Declared Resources as of 31stDecember 2017	187
Table14-60 Yalea 2017/2016 Mineral Resource Comparison	188
Table14-61 Gara 2017/2016 Mineral Resource Comparison	189
Table14-62 Loulo 3 2017/2016 Mineral Resource Comparison	189
Table14-63 Baboto 2017/2016 Mineral Resource Comparison	190
Table14-64 Gara West 2017/2016 Mineral Resource Comparison	190
Table14-65 Gounkoto Open Pit 2017/2016 Mineral Resource Comparison	191
Table14-66 Gounkoto Underground 2017/2016 Mineral Resource Comparison	192
Table14-67 Faraba 2017/2016 Mineral Resource Comparison	192
Table14-68 Loulo Gounkoto MCF 2017 EOY Reconciliation (Old assign density LEFT and new estimated density + domain assign RIGHT)	193
Table15-1 Loulo Gold Mine Ore Reserve Statement as of 31stDecember 2017	198
Table15-2 Gounkoto Gold Mine Ore Reserve Statement as of 31stDecember 2017	199
Table15-3 Loulo Open Pit Ore Reserve as of 31stDecember 2017	201
Table15-4 Loulo Surface Stockpile Ore Reserve as of 31stDecember 2017	201
Table15-5 Gounkoto Open Pit Ore Reserve as of 31stDecember 2017	202
Table15-6 Loulo Underground Ore Reserves as of 31stDecember 2017	203
Table15-7 Gounkoto Underground Project – Total Ore Reserve by Category as of 31st December 2017	203
Table15-8 Baboto – Loulo 3 - Gara West - Slope Parameters	205

18^th September 2018 Page ix

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-9 Geotechnical Slope Requirements	206
Table15-10 Basic Geotechnical Parameters for Yalea and Gara	207
Table15-11 Basic Geotechnical Parameters for Gounkoto	208
Table15-12 Summary of Actual 2017 Loulo Underground Mining Dilution	211
Table15-13 Summary of Loulo Underground Ore Reserve Estimate Dilution	212
Table15-14 Summary of Loulo Underground Ore Reserve Estimate Dilution	212
Table15-15 Summary of Loulo Underground Ore Reserve Estimate Dilution	212
Table15-16 Gounkoto Underground Mining Dilution and Losses	213
Table15-17: Loulo Open Pit Economic Parameters	214
Table15-18: Gounkoto Open Pit Economic Parameters	216
Table15-19 Yalea and Gara Underground Mine –Cut-Off Grade Calculation	218
Table15-20 Gounkoto Underground Mine –Cut-Off Grade Calculation	219
Table15-21 Whittle Optimisation Results for Loulo 3	220
Table15-22 Whittle Optimisation Results for Baboto	221
Table15-23 Pit by Pit Graph of Whittle Optimisation Results for Gara West	222
Table15-24 Whittle Optimisation results for Gounkoto	224
Table15-25 Whittle Optimisation Results for Faraba	224
Table15-26 Open Pit Design Characteristics	225
Table15-27 MSO Parameters for MSO Run	230
Table15-28 Loulo Open Pits Ore Reserve Reconciliation	232
Table15-29: Gounkoto Open Pit Ore Reserve Sources Reconciliation	232
Table16-1 Baboto Main Mining Equipment	234
Table16-2 Loulo Open Pit Life of Mine Plan	237
Table16-3 Gounkoto Mining Fleet as of 31stDecember 2017	239
Table16-4 Life of Mine Production Plan for the Gounkoto Pit	241
Table16-5 Yalea and Gara Mining Fleet as of 31st December 2017	243
Table16-6 Yalea Life of Mine Production Schedule	250
Table16-7 Gara Life of Mine Production Schedule	251
Table16-8 Productivity Rate	260
Table16-9 Time Rates for Activities	261
Table16-10 Gounkoto Production Schedule and Key Mine Physicals	261
Table16-11 Loulo-Gounkoto Life of Mine Plan	264
Table17-1 Processing Plant Production Statistics	267
Table17-2 Loulo Processing Plant Production History	272
Table17-3 Actual Process and Plant Engineering Operating Costs for 2016 and 2017	274
Table21-1 LOM Capital Expenditure	293
Table21-2 Loulo Capital Cost	295
Table21-3 Gounkoto Capital Cost	295

18^th September 2018 Page x

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table21-4 LOM Operating Unit Costs for Loulo and Gounkoto	296
Table21-5 LOM Operating Total Costs for Loulo and Gounkoto	297
Table25-1 Loulo Risk Rating	307
Table25-2 Gounkoto Risk Rating	309

18^th September 2018 Page xi

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

List of Figures

Figure1-1 Simplified Flow Sheet	14
Figure1-2 Loulo-Gounkoto Processing Milled Tonnes by Ore Source	15
Figure4-1 Location of Loulo-Gounkoto within Mali	37
Figure4-2 Loulo and Gounkoto Permit Areas	40
Figure5-1 Loulo Mine Site Layout	45
Figure5-2 Gounkoto Mine Site Layout	46
Figure6-1 Loulo Mineral Resource and Ore Reserve Evolution	51
Figure6-2 Gounkoto Mineral Resource and Ore Reserve Evolution	51
Figure7-1 Location of the Kedougou-Kenieba-Inlier in the West African Craton	53
Figure7-2 Loulo Gounkoto Regional Interpreted Geology	55
Figure7-3 Loulo-Gounkoto Permit Targets and Geology	57
Figure7-4 Yalea Simplified Geology Cross Section	58
Figure7-5 Early pink Albite ± Carbonate Alteration. B. Early Light Brown Albite±-Carbonate
Alteration and Narrow Bands of Sericite ±Chlorite	58
Figure7-6 Long Section Looking West Showing Structural Model for Yalea Deposit	60
Figure7-7 Cross Section of Gara Fold Structures	61
Figure7-8 Typical Plan View and West-East Cross Section across Gounkoto MZ1, MZ2 and MZ3	65
Figure 11- 1 Summary of Diamond Core Sample Preparation Flowchart - Exploration and Grade Control	81
Figure11-2 Summary of RC, Channel, and Trench Sample Preparation Flowchart - Exploration and Grade Control	82
Figure11-3 SGS Loulo Laboratory – Summary of Sample Preparation Procedure	83
Figure11-4 SGS Loulo Laboratory – Summary of Fire Assay (FAA505) Procedure	85
Figure11-5 Tram Line of Gara CRM’s Assayed by SGS Loulo During the Reporting Period	89
Figure11-6 Tram Line of Yalea CRM’s Assayed by SGS Loulo During the Reporting Period	90
Figure11-7 Tram line Plot of all Gounkoto CRMs Assayed by SGS Loulo	92
Figure11-8 Log Scatter Plot of Gara Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period	95
Figure11-9 HARD Plot of Gara Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period	95
Figure11-10 Log Scatter Plot of Yalea Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period	96
Figure11-11 HARD Plot of Yalea Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period	96
Figure11-12 Log Correlational Plot of Gounkoto Field Duplicate Assayed by SGS Loulo	97

18^th September 2018 Page xii

Loulo-Gounkoto Gold Mine Complex NI43-101 Technical Report

Figure11-13 HARD Plot of Gounkoto Field Duplicates Assayed by SGS Loulo	98
Figure11-14 Log Scatter Plot of Yalea Umpire Assays	99
Figure11-15 Log Scatter Plot of Gara Umpire Assays	100
Figure11-16 Log Scatter Plot of Baboto Umpire Assays	100
Figure11-17 Log Scatter Plot of Gounkoto Umpire Assays	101
Figure12-1 The Normalised Assay Structure Utilised by DataShed	105
Figure13-1 Gounkoto Cyanidation Sample Locations	109
Figure13-2 Yalea Copper Estimation within 2017 Resource Model Looking East	111
Figure13-3 Yalea Arsenic Estimation within 2017 Resource Model Looking East	111
Figure13-4 Process Recovery Estimated into Yalea Block Model	112
Figure13-5 Loulo Processing Plant Overall Recovery in 2016 and 2017	113
Figure13-6 Treatment Regime in the Intermediate Plant	114
Figure13-7 Effective Reduction in Tenor	115
Figure14-1 Yalea Model Geological Domains (Looking East). A Portion of 9007 is Also Referred to as Yalea South HG Plunge	121
Figure14-2 ‘Purple Patch’ Boundary Update Long Section	122
Figure14-3 Gara Model and Geological Domains	123
Figure14-4 Loulo 3 Model Geological Domains	124
Figure14-5 Baboto Model Geological Domains	125
Figure14-6 Gara West Model Geological Domains	126
Figure14-7 Gounkoto Model Geological Domains	127
Figure14-8 Faraba Model Geological Domains	128
Figure14-9 Yalea 2017 ‘Purple Patch’ Density Distribution	131
Figure14-10 Yalea 2017 ‘Purple Patch’ Model and Density Data Spatial Distribution Highlighting Lower Domain Contact	132
Figure14-11 Yalea 2017 Mineral Resource Block Model Densities	132
Figure14-12 Yalea 9001 Grade Distribution (Left: Log Histogram & Right: Log Probability)	139
Figure14-13 Yalea South Variogram Including 9001 and 9002 Domains	143
Figure14-14 Gara Variogram for the 3000 Domain which was Applied to All Domains with the Bearing, Plunge and Dip Changing Based on Domain Orientation	144
Figure14-15 Loulo 3 Main Zone 2 Individual Structure Normal Score Variogram Models	145
Figure14-16 Loulo 3 Main Zone 2 Nested Back Transformed Variogram Model	146
Figure14-17 Baboto South Variogram	147
Figure14-18 Baboto South 2017 Nested Back Transformed Variogram Model	147
Figure14-19 Gara West Domain 20000 Individual Structure Normal Score Variogram Models	148
Figure14-20 Gara West Domain 20000 Nested Back Transformed Variogram Model	149
Figure14-21 Modelled Semi-Variograms for Gounkoto MZ1 Domain 1000 (East Dipping)	150

18^th September 2018 Page xiii

Loulo-Gounkoto Gold Mine Complex NI43-101 Technical Report

Figure14-22 Gounkoto MZ1 Domain 1000 Nested Back Transformed Variogram Model	150
Figure14-23 Faraba Domain 1000 Modelled Semi-Variograms	151
Figure14-24 Faraba Domain 1000 Nested Back Transformed Variogram Model	151
Figure14-25 QKNA Results for Gounkoto MZ1 East Dipping Grade Control Area	152
Figure14-26 Yalea 2017 Drill hole Density Sub Domains	160
Figure14-27 Gara 2017 Estimation Domains	162
Figure14-28 Baboto Estimation Domains Looking WNW	165
Figure14-29 Gara West EstimationSub-Domains	167
Figure14-30 Yalea Mineral Resources Classification with Estimation Composites	171
Figure14-31 Gara Mineral Resources Classification with Estimation Composites	171
Figure14-32 Loulo 3 Mineral Resource Classification with Estimation Composites	172
Figure14-33 Baboto 2017 Mineral Resource Classification Surfaces with Estimation Composites	172
Figure14-34 Gara West 2017 Mineral Resource Classification Surfaces with Estimation Composites	172
Figure14-35 Gounkoto July 2017 MZ1 Classification	173
Figure14-36 Faraba Classification Within Pit	174
Figure14-37 Swath Plot of Yalea Grade Control Area of 9006	178
Figure14-38 Swath Plot Along Y Axis of 3101 (3000 GC Area)	178
Figure14-39 Yalea 2017 Resource Model Looking East Regularised for Presentation Purposed and Composite Samples Overlain onto Display Relative Accuracy of the Estimate	179
Figure14-40 Gara 2017 Resource Model Looking East Regularised for Presentation Purposed and Composite Samples Overlain onto Display Relative Accuracy of the Estimate	179
Figure15-1 Sketch of Baboto Dewatering Arrangement	209
Figure15-2 Pit by Pit Graph of Whittle Optimisation Results for Loulo 3	220
Figure15-3 Pit by Pit Graph of Whittle Optimisation Results for Baboto	221
Figure15-4 Whittle Optimisation Results for Gara West	222
Figure15-5 Gounkoto Pit Design Oblique View looking upwards from underneath, with Gain in Ore Tonnes	223
Figure15-6 Whittle Optimisation Results for Gounkoto	223
Figure 15-7 Whittle Optimisation Results for Faraba	224
Figure15-8 Baboto Pit and Waste Dump Design	226
Figure15-9 Gara West Pit And Waste Dump Design	227
Figure15-10 Loulo 3 Pit Design	227
Figure15-11 Yalea Mine Design with Ore Classification	229
Figure15-12 Gara Mine Design with Ore Classification	229
Figure15-13 Longitudinal Projection Looking West showing Results of MSO Run on Northern End of the Resource Model	231

18^th September 2018 Page xiv

Loulo-Gounkoto Gold Mine Complex NI43-101 Technical Report

Figure16-1 Baboto Mine Design	235
Figure16-2 Baboto Haul Road	236
Figure16-3 Yalea Life of Mine Plan Development Long Section (Looking East)	244
Figure16-4 Gara Life of Mine Plan Development Long Section (Looking West)	244
Figure16-5 Yalea Stoping Method Long Section	245
Figure16-6 Gara Stoping Method Long Section	245
Figure16-7 Stoping Under Rock Fill (SURF) Development Layout at Top of Mining Area (85 Level), Showing Waste Tipping Points	246
Figure16-8: Stoping Under Rock Fill (SURF) Development Layout of Production Level (185 Level)	247
Figure16-9 Isometric View of Underground Design in Relation to Super Pit Design	252
Figure16-10: Longitudinal Projection Looking East Showing Decline and Level Development	253
Figure16-11 View Looking East of Typical Level Designs Showing Variation in Position of Level	254
Figure16-12 Plan View of Typical Level Layout	254
Figure16-13 Longitudinal Projection Looking East Showing Stopes at an Average Au Grade	256
Figure16-14 Longitudinal Projection Looking East Showing Stope Blocks by Mining Type	256
Figure16-15 Longitudinal Projection Looking West Showing Stope Blocks and Stope Types	257
Figure16-16 Longitudinal Projection of Transverse Stope Layout	258
Figure16-17 Longitudinal Projection of Longitudinal Longhole Stoping	259
Figure16-18 Development Coloured by Year	262
Figure16-19 Stoping Coloured by Year	262
Figure17-1 Simplified Metallurgical Flowsheet	266
Figure17-2 Loulo-Gounkoto Processing Milled Tonnes by Ore Source	272
Figure18-1 Loulo Major Infrastructure Locations	275
Figure18-2 Starter Wall Built on East Side of TSF	279

18^th September 2018 Page xv

Loulo-Gounkoto Gold Mine Complex NI43-101 Technical Report

1 Executive Summary

The purpose of this report is to support the public disclosure of the year end 2017 Mineral Resource and Ore Reserve estimates at the Loulo-Gounkoto Gold Mine Complex (Loulo-Gounkoto, the Mine or the Project), consisting of the Loulo Gold Mine (Loulo) and the Gounkoto Gold Mine (Gounkoto) located in western Mali. This Technical Report conforms to National Instrument43-101- Standards of Disclosure for Mineral Projects (NI43-101). All currency in this report is US dollars ($) unless otherwise noted.

Société des Mines de Loulo SA is an exploration and mining company and the owner of the Loulo mine, which is held 80% by Randgold Resources Limited, (Randgold) and 20% by the state of Mali.

Société des Mines de Gounkoto SA is an exploration and mining company and the owner of the Gounkoto mine which is held 80% by Randgold and 20% by the state of Mali.

Randgold is the operator of Loulo and Gounkoto. Randgold is developing, and operating gold mines in West and East Africa. The most notable of these are the following:

● Morila Gold Mine in Mali.

● Tongon Gold Mine in Côte d’Ivoire.

● Kibali Project in Democratic Republic of Congo (DRC).

● Massawa Exploration Project in Senegal.

The Project is an operating mine site comprising two underground mines (Yalea and Gara) at Loulo, the Gounkoto open pit mine, satellite deposits, and a processing plant (4.8 Mtpa capacity), together with other associated mine operation and regional exploration infrastructure. The plant produces gold doré bars.

Total mine production from underground and open pits in 2017 was 4.9 Mt at a head grade of 5.0 g/t Au for a total of 730 koz of gold (92.7% recovery).

Total production since mining commenced in 2005 to year end 2017 is 391 Mt (53 Mt ore) for 5.6 Moz of gold

(90.2% recovery).

The effective date of this report is 31^st December 2017.

1.1 Property Description and Location

Loulo-Gounkoto is situated in western Mali adjacent to the Falémé River which forms the international boundary with Senegal. The Republic of Mali is a landlocked country and is bordered by Guinea, Senegal, Mauritania, Algeria, Niger, Burkina Faso, and the Cote d’Ivoire.

18^th September 2018 Page 1

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The Project area is located 350 km west of the capital city of Bamako, 220 km south of Kayes and to the NW of the nearest town Kenieba. It falls within the Central Arrondissement of the Kenieba District which is one of the 10 districts of the Kayes Region.

The Dakar to Bamako Millennium highway crosses the Loulo-Gounkoto haul road, approximately 6 km north of the Gounkoto pit. This highway serves as the primary access point for the mine and provide excellent road transport links with the rest of the country as well as to Senegal for which the border is within 3 km of the mine.

A 1.5 km laterite airstrip is present within the Loulo Permit which can accommodate small to medium sized aircraft and has been awarded full certification by the transport authorities in Mali. Charter flights are arranged from the site to the capital Bamako, which is served daily by commercial flights from European cities.

The Loulo and Gounkoto areas are currently 261.23 km² and 99.95 km² respectively, for a total area of 361.18 km². The Project falls within the Central Arrondissement of the Kenieba District which is one of the 10 districts of the Kayes Region.

Loulo consists of multiple mineral deposits including; Yalea, Gara, Loulo 3, Baboto, Gara West, P129, P125L3, P129QT, Loulo 1, Loulo 2 andL2-L3 Gap, P125L3, and PQ10.

Gounkoto consist of multiple mineral deposits including; Gounkoto and Faraba.

1.2 Mineral Rights and Land Ownership

The Loulo mine is within the Loulo Exploitation Permit (Loulo Permit). The original Loulo Permit was granted by Decret No.96-048/PM-RM 9, on the 14^th February 1996 and covered an area of 48 km². This Permit was amended by Decret No.99-193/PM-RM dated 15^th July 1999 and extended the size of the Loulo Permit to 372 km². The Loulo Permit was amended by DecretNo. 2012-311/P-RM on 21^st June 2012 which reduced the size of the Lolo Permit as the portion surrounding the Gounkoto Mine was transferred to a new Exploitation Permit. The Loulo Permit, that covers the Gara and Yalea underground Ore Reserves and the Baboto, Gara West and Loulo 3 open pit Ore Reserves, remains in force for a period of 30 years after which it is renewable if production is still taking place.

In 2010, Randgold applied and was granted the new Gounkoto Exploitation Permit (Gounkoto Permit), which was split from the Loulo Permit, to form a separate entity, Société des Mines de Gounkoto SA (Gounkoto) under DecretNo.2012-431/PM-RM DU dated 3^rd August 2012. The Gounkoto Permit, that incorporates the Gounkoto and Faraba Reserves, is valid for 30 years.

In 2017, the Baboto North deposit was purchased by Endeavour Mining Corporation. This resulted in a minor change to the Loulo Permit. An updated Decret number has not yet been generated by the Malian Government, but it should be created imminently.

The Loulo-Gounkoto Establishment Convention regulates the fiscal conditions under which the Loulo and Gounkoto mined operates and is based on the1991 Mining Code. A 6% royalty is

18^th September 2018 Page 2

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

payable to the Malian government based upon production together with a corporate tax rate on profits at 305 and a minimum of 0.75% on gross revenues if a loss is made. Loulo received a received a five year tax holiday from first commercial gold production in October 2005. Gounkoto received a two year tax holiday from first gold production in 2013 and has since received governmental approval for use of 50% corporate tax reduction for the next four years to support its development of a super pit. The convention includes exoneration on fuel duties for the life of the project and on import duties for three years from the start of first gold.

In the Qualified person’s opinion, all appropriate permits have been acquired and obtained to conduct the work proposed for the property.

The Qualified Person is not aware of any risks that could result in the loss of ownership of the deposits or loss of the permits, in part or in whole.

The Qualified Person is not aware of any other significant factors and risks that may affect access, title, or the right of ability to perform work on the property.

1.3 History

Gold potential was first recognised by Syndicat Or joint venture between the Malian Direction Nationale de la Géologie et des Mines and the French Bureau de Recherches Géologiques et Minières (BRGM). The Gara gold deposit was discovered in 1981 by the Syndicat Or joint venture. In 1992, BHP Minerals Mali entered into an agreement with Société des Mines de Loulo for a joint venture that developed the Gara deposit into a Mineral Resource that was deemedsub-economic at the time.

During 1996, Randgold acquired BHP Minerals Mali and undertook additional regional exploration which resulted in the discovery of Yalea, the second of two deposits that make up the Loulo Mine, in 1997. In 2003, a feasibility study was undertaken at Loulo on a 12 Mt at 3.60 g/t Au for 1.4 Moz Ore Reserve which led to the construction of an open pit mine in 2004. The Loulo underground mine passed through feasibility study stage in 2005, with development beginning in 2006 and first gold being produced in 2005. The first gold from Gara underground was produced in 2011. Someopen-pit near surface oxide mining of soft material was undertaken at some of the minor satellite deposits at various times to feed the plant as required.

Gounkoto was discovered through regional exploration in 2009 with first gold being produced at the Gounkoto open pit in 2011. As a result of the discovery of Gounkoto, in 2010 Randgold applied for and was granted the formation of a new exploitation licence that covered the southern half of the former Loulo permit and included the Gounkoto and Faraba deposits. During 2016, an independent undergroundpre-feasibility study was completed as well as a newopen-pit design for the Gounkoto‘Super-Pit’ which incorporated significant push backs and deepening that converted some underground resources and reserves intoopen-pit resources and reserves.

18^th September 2018 Page 3

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

1.4 Geology and Mineralisation

Loulo-Gounkoto is located within the Kedougou-Kenieba erosional inlier. The inlier is unconformably overlain by Upper Proterozoic sandstones towards the east and further south. The Kenieba inlier contains several significant gold deposits including Sadiola, Yalea, Segala, Tabakoto, and Gounkoto deposits in Mali, and Sabodala in Senegal. The Senegal-Mali shear marks a major break in the geology from shelf carbonates with the Falémé ironstone unit in the west to the sedimentary sequences of the Kofi formation in the east. Loulo-Gounkoto is predominately underlain by the Kofi formation consisting of greywacke, sandstone, argillaceous sandstone, calcareous sandstone and tourmalinised sandstone, sheared greenstone units. This geological setting is the primary host of mines in Burkina Faso, Ghana, Mali, Niger, and Senegal. These deposits tend to have significant strike and depth potential, with exploration concentrating on delineating strike and depth extent, followed by infill drilling within the zones of better continuity and grade.

At Yalea, the main mineralised body is a hosted by the Yalea Shear, where it is intercepted by the Yalea Structure. The Yalea Shear is a brittle-ductile, north-south striking, mineralised fault that transects the Yalea Structure, which is a complex, north to NNE striking shear zones. The Yalea mineralisation is predominantly hosted in hydrothermally brecciated argillaceous pink quartzites situated. A higher grade ‘Purple Patch’ zone is observed in a dilatational strain transfer zone formed as the western dip of the upper mineralised system steepens, forming hydraulic breccias. Economic levels of gold mineralisation are almost exclusively associated with paragenetically late sulphide veins, breccias and zones of massive sulphides. Higher grade material commonly contains sulphide veins which cut the various generations of albite ± carbonate alteration. There is a strong correlation between sulphide intensity and gold mineralisation with the dominant sulphide phases consisting of pyrite (abundant), arsenopyrite and minor chalcopyrite. The sulphides can be disseminated or massive along fabric.

Yalea mineralisation, remains open at depth and to the south with potential for significant high-grade extensions. To the south of the ‘Purple Patch’ zone the Yalea Shear forms asub-horizontally plunging transfer zone where the competency contrast of the footwall argillaceous quartzites (SQR) contact causes a transfer of strain associated with normal movement. Consequently, this transfer creates dilatational hydraulic breccias with the potential to host significant high-grade extensions.

Gara (previously known as Loulo 0) is hosted within an intensely tourmaline greywacke unit which outcrops on surface due to its high resistance to weathering. The geometry of the mineralisation is subjugated by the strike slip shearing on Senegal-Mali shear. This shearing has resulted in folding, fracturing, brecciation, and subsequent development of a quartz-carbonate vein stockworks within the brittle-ductile tourmaline altered greywacke forming what has been termed quartz tourmaline (QT) unit. On a regional scale the Gara deposit spans the hinge of a broad open fold with a gently-plunging north-south trending axis. On the deposit scale the upper limb of this fold dips has a westerly dip, whereas the lower limb dips east.

18^th September 2018 Page 4

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The distribution of gold grade in long section reflects the varying degrees of fracturing, brecciation, and subsequent development of a quartz-carbonate vein stockworks from multiple generations of folding. Gold mineralisation is strata-bound and hosted predominantly within the quartz-tourmaline stockwork veins, which are enveloped within footwall greywackes and hanging wall sandstone. Higher gold grades values typically occur where the intensity of tourmalisation and stockwork veining are strongest. The sulphide assemblages predominantly consists of disseminated auriferous pyrite with minor chalcopyrite, scheelite, and nickeliferous sulphides.

In the open pit area, the high-grade mineralisation is concentrated along thesub-horizontal fold hinge axes, whereas within the underground area, high-grade mineralisation plunges shallowly southward, parallel to the large scale open warp fold axis . These differential orientations of mineralisation are a result of the earlier, deposit scale warping locally influencing the geometry of the superimposed “S folds” during their formation.

Baboto is a shear hosted deposit situated along a north-south striking shear structure located approximately 14 km NNE from the Yalea deposit. Baboto is dominated by a thick sequence of metasediments and structural breccias. The main shear zones are vertical to steeply west dipping at Baboto South and sub vertical in Baboto Centre. Gold mineralisation is mainly associated with the finely disseminated pyrite occurring in the brittle-ductile shear breccias, which generally have a lensoidal shape defined by a series ofsub-parallelN-S shears that follow key lithological contacts.

Loulo 3 is located 4 km NNE of the Yalea mine. Loulo 3 consists of three mineralised zones: a NNW trending main zone (MZ1) which is situated on the Loulo 3 structure and is transected by the NNE striking main zone (MZ2), which is situated on the Yalea structure, and the third small sub parallel NW striking footwall zone. Mineralisation consists of a mixture of quartz and hematite veinlets hosted in a zone of silica-carbonate alteration within local tourmaline alteration in the south. The distribution of high-grade zones is controlled by the narrowing of the host stratigraphy package, which focusses strain and fluid flow, causing the hematite rich Yalea Structure to interact with the silica-carbonate Loulo 3 Structure particularly within MZ2. Gold bearing sulphides predominantly consist of pyrite and arsenopyrite, with chalcopyrite occurring as a latenon-gold bearing phase. Gara West is located 200 m west of the pit at Gara and is characterised by predominantly shear and breccia hosted mineralisation within a medium to coarse grained sandstone unit that is variably altered with tourmaline, chlorite, and silica-carbonate. The sandstone hosts four mineralised lodes striking NNE and dipping moderately westward. The gold mineralisation is strata bound as it has been preferentially altered with tourmaline (and silica-albite), due to the increased porosity of the protolith, relative to the bounding limestone in both the hanging wall and footwall.

Other minor satellite deposits are present within the Loulo Permit, these exhibit similar geological characteristics to the other major deposits outlined above.

Gounkoto is a large NNW trending shear zone, with a complex assemblage of ductile shear breccias, shears, and faults characterised by a stepped geometry, with wider zones of mineralisation generally seen on the NW trending structures and narrower zones on the north-south trending structures. This is believed to be related to dilation across these structures in a

18^th September 2018 Page 5

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

strong sinistral strain environment. The mineralisation is generally hosted in a siliceous ‘Rose Quartzite’ (QR) unit. The mineralisation is subdivided based on the structural and lithological characteristics. From north to south, these are:

● Northern Zone: a narrow package of limestone-hosted mineralisation between hanging wall and footwall shear zones striking NNE and dipping steeply to the east. Mineralisation is of intermediate thickness and grade relative to the rest of the deposit.

● Jog Zone: a broad assemblage of repeated stacked gritty QR units, structurally offset from one another creating a stepped geometry, generally striking NNW. Upright flower structures have created an apparent thickening of units, forming wide mineralised zones of high-grade.

● Pinch Zone: a narrow package ofQR-hosted mineralisation between hanging wall and footwall shear zones striking NS and dipping steeply east close to the surface but shallows in dip at depth. The mineralisation is generally narrow andlow-grade.

● Wrench Zone: a broad package ofQR-hosted strongly mineralised lodes between hanging wall and footwall shear zones striking NW and dipping 40° to 50° east.

● Southern Zone: a narrow package ofQR-hosted mineralisation between hanging wall and footwall shear zones striking NS and dipping steeply east. The mineralisation is generally narrower and lower grade than the Northern and Wrench Zones.

The Faraba deposit strikes NNW and is comprised of several zones of gold mineralisation hosted within and along the contacts of north-south striking, coarse grained, gritty sandstone units (lithic wackes) in a package of sheared argillaceous sediments. Lithologic layering (transposed bedding) dips steeply westward; however, the mineralised zones dip steeply to the east. The mineralisation terminates where the Faraba Structure meets the argillite units on either side of the sandstones. The resulting mineralisation occurs as numerous silica-carbonate and secondary iron oxide altered sub vertical panels with narrow east-west dimension, each containingsub-horizontal to shallow plunging zones of higher grade. Gold mineralisation is dominantly hosted by pyrite, with local magnetite, chalcopyrite, arsenopyrite and pyrrhotite.

1.5 Exploration

The Project has been explored by Syndicat Or and the BRGM and more recently by BHP. Since Randgold’s acquisition of the Permit in 1996, significant exploration has been undertaken to develop both brownfields and greenfields targets. Sampling has primarily been undertaken through reverse circulation (RC) drilling, along with diamond drilling and trenching. Rotary air blasted (RAB) drilling has also been undertaken on some early stage exploration targets, although RAB drilling is not included in the Mineral Resource estimate.

Since 1993 the following sampling has been undertaken at Loulo:

● Diamond Drilling – 3,035 drill holes for 647,663 m

● RC – 10,461 drill holes for 414,012 m

● RAB – 4,438 drill holes for 103,720 m

● Trenches – 931 cuts for 49,241 m

● Underground Channels – 5,067 channels for 35,825 m

18^th September 2018 Page 6

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

^○  Total Sampling – 1,250,460 m

Exploration at Loulo-Gounkoto is focussed on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit extensions of the current Mineral Resources for high-grade mineralisation based on the structural model. The current exploration concept has been proven to be effective, with both the discovery of Gounkoto and the successful replenishment of depleted Mineral Resources and Ore Reserves at both mines.

1.6 Mineral Resources

The Mineral Resource estimates have been prepared according to the guidelines of the Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI43-101 and there are no material differences.

Quality assurance and quality control (QA/QC) has been undertaken across the life of exploration to minimise errors. A standard operating procedure (SOP) outlines Randgold’s approach to QA/QC which meets industry best practice. The results from the 2017 QA/QC program show that the performance of the SGS Loulo laboratory is meeting industry standards and the resulting data is suitable for use within a Mineral Resource estimate.

Thecut-off grade selected for limiting each of the Mineral Resources corresponds to the insitu marginalcut-off grade using a gold price of $1,500/oz.

For the open pit Mineral Resources, the pit shell selected for limiting of each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground panels were reviewed and those that were deemed as having a reasonable prospect of eventual economic extraction were included in the reported Mineral Resource.

The Qualified Person is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, fiscal, metallurgical, or other relevant factors, which could materially affect the Mineral Resource estimate.

Loulo

For the year end 2017 Mineral Resource update the Yalea and Gara underground resource models have been updated due to a significant quantity of new data from grade control and surface drilling. Additionally, open pit resources of Baboto open pit designs have been updated,

18^th September 2018 Page 7

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

with new geological models and review of estimation methodology based on additional drilling. There has been no change to other satellite projects such as Loulo 3 and Gara West since the Mineral Resources declared in year end 2016.

As of 31^st December 2017 (100% basis) the Measured and Indicated Mineral Resources for the Loulo Gold Mine are estimated to be 53 Mt at 4.64 g/t Au for 7.9 Moz of gold, with an additional Inferred Mineral Resource of 12 Mt at 3.9 g/t Au for 1.6 Moz of gold (Table1-1). These Mineral Resources have been depleted to 31^stDecember 2017 using themined-out surfaces and voids.

Table1-1 Mineral Resource Statement for the Loulo Gold Mine as of 31^st December 2017

Source	Mineral Resource	Tonnes (Mt)	Grade (g/t)	Ounces (Moz)	*Attributable Gold (Moz)
Stockpiles	Measured	1.7	1.60	0.086	0.068
Open Pit	Measured	1.9	2.67	0.17	0.13
	Indicated	6.9	3.08	0.69	0.55
	Inferred	2.5	3.3	0.27	0.21
Underground	Measured	17	4.99	2.7	2.1
	Indicated	26	5.18	4.3	3.4
	Inferred	9.8	4.1	1.3	1.0
Total	Measured	20	4.49	2.9	2.3
	Indicated	33	4.73	5.0	4.0
	Measured + Indicated	53	4.64	7.9	6.3
	Inferred	12	3.9	1.6	1.3

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in Loulo Gold Mine.

Mineral Resources are reported on an 100% and attributable basis.

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resource tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are those within a $1,500/oz pit shell at an averagecut-off grade of 0.7 g/t Au.

Underground resources are those below the $1,500/oz pit shell at acut-off grade of 1.89 g/t Au at Gara and 2.04 g/t Au at Yalea.

Mineral Resources for Loulo were generated by Mr Simon Bottoms, CGeol, an officer of the company and Qualified Person.

Numbers may not add due to rounding.

Resource definition drilling in Gara Far South extension to the south of the Gara Far South underground resource, has converted previously Inferred Resources to Indicated classification. Additional resource drilling in Yalea South has also upgraded previously Inferred Resources to Indicated classification.

Gounkoto

For the year end 2017 Mineral Resource update the Gounkoto open pit and underground models have been updated due to a significant quantity of new data from grade control and surface drilling. There has been no change to the Faraba open pit resource as no additional exploration was undertaken on the deposit.

As of 31st December 2017 (100% basis) Gounkoto Measured and Indicated Mineral Resources are estimated to be 28 Mt at 4.15 g/t Au for 3.7 Moz, with an additional 4.0 Mt at 3.1 g/t Au for 0.4 Moz of Inferred material (Table1-2). The Mineral Resources consist of Gounkoto open pit and underground Mineral Resources, surface stockpiles, and Faraba open pit. These Mineral Resources have been depleted to 31st December 2017 using themined-out surfaces and voids.

18^th September 2018 Page 8

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table1-2 Mineral Resource Statement for the Gounkoto Gold Mine as of 31^st December 2017

Source	Mineral Resource	Tonnes (Mt)	Grade (g/t)	Ounces (Moz)	*Attributable Gold (Moz)
Stockpiles	Measured	1.8	1.96	0.11	0.089
Open Pit	Measured	5.4	4.33	0.75	0.60
	Indicated	18	4.04	2.3	1.9
	Inferred	1.4	2.3	0.11	0.08
Underground	Indicated	3.0	5.74	0.56	0.45
	Inferred	2.6	3.5	0.29	0.23
Total	Measured	7.1	3.75	0.86	0.69
	Indicated	21	4.28	2.9	2.3
	Measured + Indicated	28	4.15	3.7	3.0
	Inferred	4.0	3.1	0.40	0.32

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in Gounkoto Gold Mine. Mineral Resources are reported on an 100% and attributable basis.

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resource tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are those within a $1,500/oz pit shell at an averagecut-off grade of 0.8 g/t Au.

Underground Mineral Resources are those below the $1,500/oz pit shell at acut-off grade of 2.0 g/t Au.

Mineral Resources for Gounkoto were generated by Mr Simon Bottoms, CGeol, an officer of the company and Qualified Person.

Numbers may not add due to rounding.

A decrease of approximately 200 koz resources from the year end 2016 Mineral Resource estimate to the 2017 Resource Estimate is directly attributable to the 2017 open pit mining depletion which was partially replenished through positive model changes defined by grade control drilling in 2017.

1.7 Ore Reserves

The Ore Reserve estimates have been prepared according to the guidelines of the Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI43-101 and there are no material differences.

The Ore Reserve estimates use updated economic factors, the latest resource and geological models, geotechnical inputs, and the latest metallurgical updates. Some inputs were shared across all the operations during the preparation of the Ore Reserve estimates. Ore Reserves were based on the development of appropriately detailed and engineered life of mine (LOM) plans. All design and scheduling work was undertaken to a suitable level of detail by experienced engineers using mine planning software. The planning process incorporated appropriate modifying factors and the use ofcut-off grades and other technical-economic investigations. Ore Reserves are stated:

● As of 31^st December 2017

● At a gold price of $1,000/oz

● AsRun-of-Mine (ROM) grades and tonnage as delivered to the fully diluted and delivered to the plant

18^th September 2018 Page 9

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

● Including only Measured and Indicated Mineral Resources.

The average dilution for the underground deposits ranged from 10% to 25%, with mining losses estimated at 4% to 8%. The average dilution for the open pit deposits was 10% with mining losses estimated at 3% to 4%.

The Qualified Person has performed an independent verification of the block model tonnes and grade, and in their opinion, the process has been carried out to industry standards.

The Qualified Person is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, fiscal, infrastructure, permitting, that could materially affect the Ore Reserve estimate.

Loulo

The year end 2017 (100% basis) Loulo total Proved and Probable Ore Reserves are estimated to be 36 Mt at 4.51 g/t Au containing 5.2 Moz of gold of which 4.1 Moz are attributable to Randgold.

The Ore Reserves estimate for Loulo as of 31st December 2017 are detailed in Table1-3.

Table1-3 Ore Reserve Statement for the Loulo as of 31^st December 2017

Source	Ore Reserve	Tonnes   (Mt)	Grade g/t	Gold   (Moz)	*Attributable   Gold (Moz)*
Stockpiles	Proved	1.7	1.60	0.086	0.068
Open Pit	Proved	1.5	2.43	0.12	0.093
	Probable	3.9	3.87	0.48	0.39
	Proved + Probable	5.4	3.47	0.60	0.48
Underground	Proved	8.8	4.97	1.4	1.1
	Probable	20	4.82	3.1	2.5
	Proved + Probable	29	4.87	4.5	3.6
Total	Proved	12	4.18	1.6	1.3
	Probable	24	4.67	3.6	2.9
	Proved + Probable	36	4.51	5.2	4.1

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in the Loulo Gold Mine. Ore Reserves are reported on a 100% and attributable basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz and an averagecut-off grade of 1.1 g/t Au including dilution and ore loss factors.

Underground Ore Reserves are reported at a gold price of $1,000/oz and acut-off grade of 2.69 g/t Au for Yalea and 2.4 g/t Au for Gara including dilution and ore loss factors.

Open pit and underground Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

Loulo year end 2017 Ore Reserve estimate shows a net reduction of 85 koz when compared to the estimate for year end 2016. This is entirely attributed to the sale of Baboto North. The 0.54 Moz annual depletion for the year was replenished through:

● 0.18 Moz increase in the Yalea South Ore Reserves with resource conversion drilling which has resulted in the extension of the Yalea South decline to-620 mRL;

● 0.39 Moz increase in the Gara Ore Reserve from the conversion drilling to the south of Gara (Gara Far South Extended).

18^th September 2018 Page 10

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

The year end 2017 (100% basis) Gounkoto Proved and Probable Ore Reserves are estimated to be 20 Mt at 4.58 g/t Au containing 3.0 Moz of gold, of which 4.1 Moz are attributable to Randgold.

The Ore Reserves estimate for Gounkoto as of 31^st December 2017 is presented in Table1-4.

Table1-4 Ore Reserve Statement for Gounkoto as of 31^st December 2017

Source	Ore Reserve	Tonnes   (Mt)	Grade g/t	Gold   (Moz)	Attributable   Gold (Moz)*
Stockpiles	Proved	1.8	1.96	0.11	0.099
Open Pit	Proved	4.4	4.75	0.66	0.53
	Probable	12	4.63	1.8	1.4
	Proved + Probable	16	4.66	2.4	1.9
Underground	Proved	-	-	-	-
	Probable	2.2	6.09	0.42	0.34
	Proved + Probable	2.2	6.09	0.42	0.34
Total	Proved	6.1	3.95	0.78	0.62
	Probable	14	4.85	2.2	1.7
	Proved + Probable	20	4.58	3.0	2.4

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in the Gounkoto Gold Mine. Ore Reserves are reported on a 100% and attributable basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz and an averagecut-off grade of 1.1 g/t Au including dilution and ore loss factors.

Underground Ore Reserves are reported at a gold price of $1,000/oz and acut-off grade 3.0 g/t Au including dilution and ore loss factors.

Open pit Ore and underground Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

The 2017 Gounkoto complex Ore Reserve estimate is 175 koz lower than estimated in 2016 as, although there was annual depletion in 2017 of 336 koz (including stockpiles), this was partially offset by positive model changes defined by infill grade control drilling.

1.8 Mining Method

Over the LOM of Loulo-Gounkoto, a total of 52 Mt of ore at 4.73 g/t Au is expected to be produced over 15 years up to 2032. Ore supplied to the plant during this period, including stockpile changes, will be 56 Mt at an average grade of 4.55 g/t Au resulting in 7.5 Moz recovered at an average processing recovery of 92%.

The Yalea underground operation will continue until 2030 and the Gara operation until 2032, with the Loulo open pits mined from 2024 through to 2027. The Gounkoto open pit, also referred to as the ‘Super Pit’, will be mined out in 2024 with the Faraba open pit starting in 2022 and ending in 2024. The Gounkoto underground operation will commence in 2025 and will continue until 2030.

A total of 28.6 Mt of ore will be mined from the Yalea and Gara underground operations with a further 5.4 Mt mined from the Loulo 3, Baboto, and Gara West open pits, at the Loulo Operations. The Gounkoto Operations will contribute 16 Mt mined ore tonnes from the Gounkoto and Faraba open pits with the Gounkoto Underground Operation adding 2.1 Mt.

18^th September 2018 Page 11

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Both the Loulo and Gounkoto share similar mining methods across their underground and open pit operations.

In the Qualified Person’s opinion, the parameters used in the Mineral Resource to Ore Reserve conversion process are reasonable.

Open Pits

The Loulo Permit are comprised of the Baboto open pit, Loulo 3 open pit, and the Gara West open pit. Baboto oxide and free dig transition ore is scheduled to be mined in 2018, however the rest of the Baboto open pit fresh ore and the other pits will be mined from 2024 onwards, after the larger Gounkoto open pit is depleted. The production from these pits has not been extensively detailed. Randgold intends to use the same contractor that has already mined part of the Baboto open pit, so the fleet requirement is known. The Loulo 3 open pit was mined previously in 2013 and the eastern push back forms the remaining Ore Reserve. Both Loulo 3 and Gara West open pits will use the same equipment as they are timed to be mined consecutively.

Gounkoto is anon-going operation that applies selective mining. Ore is classified as either Full Grade Ore (FGO), delivered to primary crushers, or Marginal Ore (MO), which is stockpiled close to the primary crushers, and blended with the FGO to provide a consistent feed grade. The Gounkoto mining plan has been sequenced in such a way that the south pit will be mined out in advance of the north pit. This will allow the waste mined from the north pit to be backfilled into the south pit resulting in a shorter haul than the conventional route of tramming out of the pit and onto a waste dump. This, however, means excess ore must be mined in 2019 and stockpiled to ensure the plant feed requirements are maintained during the waste pushback mining in years 2020 to 2022.

Open pit mining is conducted by the contractor Gounkoto Mining Services (GMS), a local subsidiary of DTP Terrassement, using eitherfree-dig or conventional drill, blast, load, and haul methods.

The mining equipment is jointly owned by a subsidiary of Randgold and the contractor’s parent, which also operates at Randgold’s Kibali Mine in the DRC and Tongon Mine in Côte d’Ivoire.

The potential ingress of groundwater are investigated prior to mining and continually updated during the mining activities. Prior to commencement of mining, dewatering well systems are installed for all pits to lower the groundwater level and a system of dewatering trenches is established, preventing the inflow of any surface water to the active mining areas.

The upper levels of the open pits are usually in weathered material, which typically is free digging material. Once fresh (unweathered) rock is encountered, drilling and blasting is required. Emulsion explosives are supplied as adown-the-hole service by Maxam.

Free digging in the upper levels is carried out using 5 m high benches, with 10 m benches used for drilling and blasting operations. The 10 m benches containing ore are excavated in three flitches of equal height.

18^th September 2018 Page 12

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Opportunities exist to upgrade and convert the Inferred Mineral Resources within the current pits to Ore Reserves with drilling, but any Inferred Resources within pit designs are not reported as Ore Reserves.

Underground

The Yalea and Gara underground mines are currently operating and are accessed via portals located in open pits and a box cut. The declines were originally developed as twin declines. One decline containing a conveyor and the second decline is used for mobile equipment access. The lower part of the mines has been developed as single declines with truck haulage up to crushers which feed ore and waste onto the conveyors.

Three mining methods are used at the Yalea and Gara underground mines. They are long hole transverse open stoping; long hole longitudinal retreat open stoping; and stoping under rock fill (SURF). SURF mining is planned to be introduced in the weathered areas of Yalea South Upper.

Previous longitudinal stoping in the upper areas of the mines uses a panel and pillar method, where approximately 10 m long pillars were left between 50 m long panels. In 2014 paste fill plants were commissioned at both mines. The paste fill plants at each of the mines produce a mixed paste / aggregate fill. With the introduction of paste fill the lower areas of the mines are being mined using an underhand (top down) long hole stoping method which retreats to central accesses in an echelon format. Panels are 50 m long, mined as single level stopes (25 m high) and filled with cemented paste fill. The paste fill is exposed by the mining of both the panel below and the next panel on the same level. Transverse long hole open stoping is used in the wider (>15 m wide) parts of Yalea North. Transverse stopes are 15 m to 30 m along strike and mined from hanging wall to footwall.

The SURF method is planned for the portion of Yalea South Upper that has weathered transition or saprolite present in the mineralised zone, or the immediate hanging wall. The method will be used is expected to provide a more consistent production in the less competent ground.

From an operational perspective SURF is very similar to longitudinal sub level caving (SLC). The exception is that in SURF waste rock is tipped into the stope to fill mined voids and prevent caving of the hanging wall. In contrast SLC works by the hanging wall caving to fill the mined void.

A diesel trackless fleet is used for production comprising of 11 LHD’s and trucks, in addition to production drills and development drills.

Gounkoto underground is at apre-feasibility level of study. The proposed mining method for Gounkoto underground consists of longhole bench stoping with backfill. Due to the relatively small reserve tonnage of Gounkoto underground, all attempts to reduce the capital outlay have been adopted. The ore handling method is designed to be decline trucking without underground crushing. This alleviates the need for any capital excavation and underground engineering. Thein-pit portal position reduces the amount of decline development required and pushes the required start date of the decline out to 2025. Utilisation of the pit alleviates the need for a separate box cut development.

18^th September 2018 Page 13

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The MZ3 mineralisation located on the footwall of the Fault Gouge fault has been removed from the undergroundpre-feasibility schedule due to the poor ground conditions modelled in the area. This has simplified the stoping method, removing the requirement for cut and fill with only longitudinal and transverse benching methods being used depending on the thickness of the ore.

The mine design consists of four mining panels, each consisting of between three to five, 20 m inter- levels. The first level of each panel will consist of 100% high strength cemented aggregate fill (CAF) that will form a backfill sill pillar. Mining stopes are scheduled to mine bottom up from each backfill pillar level, allowing mining from multiple levels at the same time. Backfill sill pillars have been positioned at the base of high-grade areas of the geological grade model to allow for an optimal mining grade profile.

1.9 Mineral Processing

The Loulo processing plant uses a carbon in leach (CIL) gold extraction process with a throughput capacity of 4.8 Mtpa. The Loulo process plant process ore from both Loulo and Gounkoto operations. A simplified flowsheet is seen in Figure1-1, which depicts a standard processing circuit for free-milling ores including conventional circuits for comminution, milling, gravity, classification, and CIL.

Figure1-1 Simplified Flow Sheet

Forecasted gold processing recovery is based on both testwork and operational history. The Yalea and Gara metallurgical recoveries for remaining LOM have been estimated at 84.5% and 94.5% respectively. The Gounkoto Super Pit testwork and historical operations data has

18^th September 2018 Page 14

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

indicated an estimated recovery of 92.5%. Yalea recovery is impacted by the presence of arsenic and copper which reduce recovery by impacting the process of adsorption of the gold onto the CIL tanks. Arsenic and copper impurities also increase cyanide and oxygen consumption. Consequently, arsenic and copper estimations are completed as part of the Mineral Resource update in order to identify potentially low recovery areas. Gold recovery is maintained above 90% by blending the various ore sources (Yalea / Gara/ Gounkoto) to control copper and arsenic grades in the mill feed.

The current LOM has an average recovery life of mine of 92.3%. The average gold recovery in 2017 was 92.7%, an improvement from 2016 (Figure1-2).

Figure1-2 Loulo-Gounkoto Processing Milled Tonnes by Ore Source

Operating costs for the process plant and plant engineering disciplines are denoted in Table1-5. The total average annual unit cost process cost for Loulo-Gounkoto in 2017 is $17.21 per processed ore tonne.

18^th September 2018 Page 15

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table1-5 Actual Process and Plant Engineering Operating Costs for 2016 and 2017

Cost	Units	2016 Actual	2017 Actual
Fixed Cost
Consultants	$ ‘000	166	153
Contractors - Assays	$ ‘000	742	798
Contractors - Other	$ ‘000	1,217	1,358
Contractors - Oxygen	$ ‘000	2,312	2,680
Equipment Hire	$ ‘000	2,537	2,523
General Costs	$ ‘000	1,142	1,520
Gold Refining	$ ‘000	1,564	1,646
Labour	$ ‘000	2,831	3,653
Stores - Electrical & Mechanical	$ ‘000	1,666	1,626
Stores - Other	$ ‘000	5,536	5,366
Total Fixed Cost	$ ‘000	19,712	21,323
Tonnes Processed	1000	4,875	4,918
Total Fixed Cost	$/t	4.04	4.34
Variable Cost
Power	$/t	5.49	6.21
Reagents - Cyanide	$/t	1.63	1.26
Reagents - Lime	$/t	0.77	0.62
Good Issues - Caustic Soda	$/t	0.11	0.15
Good Issues - Activated Carbon	$/t	0.16	0.13
Reagents - Other	$/t	0.33	0.34
Stores - Grinding Media	$/t	1.93	1.92
Stores - Screens & Panels	$/t	0.04	0.04
Total Variable Costs	$/t	10.46	10.67
Total Process Cost	$/t	14.50	15.01
Plant Engineering	$/t	2.19	2.20
Combined Plant & Engineering	$/t	16.69	17.21

Loulo-Gounkoto has demonstrated successful operation both in terms of processing throughput and in particular with gold recovery. In the opinion of the Qualified Person, no fatal flaws have been identified with this facility and it is expected to deliver an average of 4.75 Mtpa over the LOM as per the forecast.

1.10 Project Infrastructure

Primary access for staff and plant to Loulo-Gounkoto is via the recently constructed (2011) Millennium highway that runs from Dakar, Senegal to Bamako, Mali, which acts as the primary supply chain route. The Millennium Highway crosses the Loulo to Gounkoto haul road approximately 6 km north of Gounkoto and provides excellent road connections in comparison to much of the country.

Daily flights with international air carriers are available from Dakar and Bamako. Charter flights between Bamako and the (unsealed) airstrip at the mine are available, when required. The landing strip at Loulo is approximately 1.5 km long and built from laterite material. It can accommodate moderate sized aircraft and has been awarded full certification by the transport authorities in Mali.

The climate at Loulo-Gounkoto is strongly influenced by the north and southward movement of the Inter Tropical Convergence Zone (ITCZ) which creates distinctive wet and dry seasons.

18^th September 2018 Page 16

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Although the annual evaporation thereby exceeds the annual rainfall, an excess of water is available during the peak of the wet season (July to September) to generate surface water runoff. Water is sourced for the Project from the Gara and Falémé rivers which run through the Project site. Climatic conditions do not materially affect either exploration, development, or mining operations.

The Loulo-Gounkoto mine site is an operating mine site comprising open pit and underground mines, a processing plant, satellite deposits and associated infrastructure. Previously mined open pits remain open and are used to access the underground mine. Waste dumps are situated adjacent to the open pits. The plant and offices and accommodation village are located east of the Gara pit.

The Tailings Storage Facility (TSF) is located 8 km to the east of the plant in an area with a number of natural ridges. It has been designed to maintain a minimum freeboard of 1.5 m to provide sufficient storage to contain a 1 in 50 year rainfall event over a72-hour period.

Power is generated on site using light and heavy fuel generators. Installed thermal capacity is 64 MVA with a power draw that is variable but in the region of 42 MW. Substantial infrastructure already exists at Loulo-Gounkoto as a result of the long-established open pit mining operation. This includes ore processing and tailings facilities, workshops offices, and camps. Mobile telephone services are available across the mine.

1.11 Market Studies

Financial evaluation of all Ore Reserves uses a gold price of $1,000/oz and all open pit reserves are estimated within pit designs based on a gold price of $1,000/oz. This is in line with Randgold corporate guidelines. Gold price sensitivities were run for all the pits.

Financial evaluation andcut-off grade calculation for the Loulo and Gounkoto underground Ore Reserves has been based on a gold price of $1,000/oz. This is the same value used for all previous underground Ore Reserves estimates from 31^st December 2011 onwards.

Royalties payable to the Malian government remain unchanged from completion of the feasibility in 2012. A total royalty payable to the Malian government of 6% of gold was used for the open pit and underground Ore Reserve estimates.

Loulo-Gounkoto pay income tax at a rate of 30% to the Malian government.

Gold doré produced at the mine site is shipped from site under secured conditions and sold under agreement to a refinery. Under the agreement, Randgold receives the ruling gold price on the day after dispatch, less refining and freight costs, for the gold content of the doré gold. Randgold has an agreement to sell all gold production to only one customer. The “customer” is chosen periodically on a tender basis from a selected pool of accredited refineries and international banks to ensure competitive refining and freight costs. Gold mines do not compete to sell their product given that the price is not controlled by the producers.

18^th September 2018 Page 17

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

1.12 Environmental, Permitting and Social Considerations

Investigation of environmental and social aspects of open pit and underground mining operations at Loulo and Gounkoto has been undertaken. The Environmental Impact Assessments (EIAs) concluded that no significant impacts were identified which will impact the physical, biological, or social environment. The continuing operations beyond the original life of mine contribute to extending employment of local Malians and growth of the Malian economy.

An environmental management plan (EMP) is in place, the Loulo operations are ISO 14001 compliant and independently audited to continuously improve environmental management. All environmental permits are in place for the Loulo processing plant and the Yalea / Gara underground mines and the Gounkoto Super Pit. The site is also audited against the requirements of the International Cyanide Management Code.

Waste rock is generated from the open pit and underground operations; the waste rock dump at Gounkoto is close to the Falémé River which forms the border with Senegal. The waste dump has been carefully designed to minimise seepage and a collection system is in place. Waste rock will also be disposed of into the southern portion of the Gounkoto Super Pit to minimise surface disposal.

Waste rock geochemistry has been studied and revealed potential acid rock drainage (ARD) generation and metals leaching from specific lithologies. However, this is closely monitored and no significant ARD issues have occurred.

Tailings are generated from the Loulo plant and disposed of in the tailings storage facility (TSF) some 8 km east of the plant. The relevant Environmental and Social Impact Assessment (ESIA) and Management Plan for permitting requirements are underway for the expansion of the current TSF.

Routine environmental monitoring takes place across the site, including dust deposition, noise, arsenic, and weak-acid dissociated (WAD) cyanide sampling, TSF seepage water and tails streams as well as sample collection of drinking water, ground water, surface water and the TSF borehole water.

Environmental incidents are noted in a register which forms part of the Environmental Management System (EMS); the causes and responses are identified, and once completed, the incident is closed out.

A water balance model has been developed for Loulo which models flows, inputs and losses, including the open pits and underground workings, plant, TSF, water management structures, offices, camp, and treatment facilities.

Mine closure costs are updated each year, with increases or decreases in disturbed areas noted and costed; the current cost for rehabilitation and closure of the mine according to the calculation model is $29.4 M for Loulo and an additional $9.7 M for Gounkoto.

18^th September 2018 Page 18

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Randgold continually engages with local communities and focuses on potable water supplies, primary school education, health care and education. The mine is a significant employer to members of the local communities. The underground mining operations contribute to extended life of mine, employment of local Malians and the growth of the Malian economy. Randgold’s policy is to promote nationals to manage the Project. Unskilled labour is typically sourced from the local area while more skilled posts are filled by staff from elsewhere in Mali, including Bamako.

Local procurement is also promoted and is a requirement of contractors as well as Randgold. Where possible, goods and services are procured locally. This includes produce from an agribusiness started by Randgold which is purchased for use in the mine canteens.

Stakeholder engagement activities, community development projects and local economic development initiatives contribute to the maintenance and strengthening of the Loulo-Gounkoto Social License to Operate (SLTO). A grievance mechanism is in place. No grievances were received in 2017; the last grievance was recorded in 2015. Through the stakeholder engagement process, concerns are raised by the community.

There is a significant ongoing presence of artisanal miners (orpailleurs) operating within the Loulo Permit area, particularly along the haul road. Artisanal miners or small-scale miners (ASM) represent a significant portion of local households’ livelihood. The government of Mali, as part owners of the operations, provide security in the form of Gendarmes who intervene on request if ASM activities interfere with the operation of the mine. The mining industry in Mali has established a committee to deal with the issue at the highest level of the Government.

The Qualified Persons consider the extent of all environmental liabilities to which the property is subject to have been appropriately met.

1.13 Capital Costs

Loulo-Gounkoto is anon-going combined open pit and underground mining operation with the necessary facilities, equipment, and manpower in place to produce gold.

The open pit and underground LOM and capital and operating cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

The majority of the capital cost estimates contained in this report are based on quantities generated from the open pit and underground development requirements and data provided by Randgold.

Capital expenditure over the remaining LOM is estimated to be $598M. A breakdown of the expenditure is detailed in Table1-6.

18^th September 2018 Page 19

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table1-6 LOM Capital Expenditure

Capital Expenditure	Cost ($ M)
Construction and Project Capital	1.4
Ongoing Capital	107
Underground Capital Development and Drilling	311
Pre-Production Capitalised	137
Exploration Capitalised	2.9
Rehabilitation/Mine Closure	39
Total	598

1.14 Operating Costs

Loulo-Gounkoto maintains detailed operating cost records that provides a foundation for estimating future operating costs. Costs used for the open pit optimisations were derived from the Mining Contractor’s pricing of the open pit LOM schedule.

Labour costs for national employees were based on actual costs. Local labour laws regarding hours of work, employment conditions were also considered, and overtime costs included.

During 2017 costs for processing and general and administration (G&A) were updated based on actuals adjusted with the latest forward estimates, production profiles and manning levels.

Customs duties, taxes, charges, and logistical costs have been included.

The LOM combined open pit and underground operating cost for Loulo is estimated to be $68.90/t milled and for Gounkoto it is estimated to be $68.49/t. Unit costs used to estimate LOM operating costs are summarised in Table1-7.

Table1-7 LOM Operating Costs for Loulo and Gounkoto

Activity	Units	Loulo	Gounkoto
Open Pit Mining	$/t mined	2.69	2.64
Open Pit Mining	$/ore t mined	34.10	36.95
Underground Mining	$/t mined	45.65	57.99
Underground Mining	$/ore t mined	46.35	82.40
Stockpile Movement	$/t milled	-3.63	3.81
Processing	$/t milled	18.22	17.03
Trucking & Hauling	$/t milled	0.03	5.20
G&A	$/t milled	8.30	7.73
Mining Total	$/t milled	42.35	38.53
Total LOM Net OPEX	$/t milled	68.90	68.49

1.15 Economic Analysis

This section is not required as the property is currently in production, Randgold is a producing issuer, and there is no material expansion of current production. RPA has verified the economic viability of the Ore Reserves via cash flow modelling, using the inputs discussed in this report.

18^th September 2018 Page 20

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

1.16 Interpretation and Conclusions

Geology and Mineral Resources

The Loulo and Gounkoto operation has documented standard procedures for the drilling, logging, and sampling processes which meet industry standards. The geological and mineralisation modelling at Loulo and Gounkoto is based on visibly identifiable geological contacts, which ensure a geologically robust interpretation.

The Loulo and Gounkoto has a quality control program in place to ensure the accuracy and precision of the assay results from the analytical laboratory. Checks conducted on the quality control database indicated that the results are of acceptable precision and accuracy. In the Qualified Person’s opinion, the sample selection, preparation, and analysis are suitable for use within a Mineral Resource estimate.

Geological models and subsequent Mineral Resource estimations have evolved and improved with each successive model update. Significant grade control drill programs and mapping of exposures within mine developments have been completed to increase the confidence in the Mineral Resource and Ore Reserve estimates. During the last three years, significant resource extension drilling has been undertaken at both Yalea and Gara. This has led to the addition of new extensions Yalea Far South, Gara Far South, and Gara Far South Extended, making a major impact towards the replenishment of 2016 and 2017 depletion.

In the Qualified Person’s opinion, both the Loulo Mineral Resources and Gounkoto Mineral Resources top cutting, domaining and estimation approach are appropriate and use industry accepted methods. The Qualified Person’s consider the Mineral Resources at Loulo and Gounkoto are appropriately estimated and classified.

The Qualified Person is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, fiscal, metallurgical, or other relevant factors, which could materially affect the Mineral Resource estimate.

Exploration at Loulo-Gounkoto is focussed on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit extensions of the current Mineral Resources for high-grade mineralisation based on the structural model. During 2017, an updated tectonostratigraphy of the Kofi Series rocks at Loulo-Gounkoto has been developed, to improve the model of regional geologic architecture and understanding of key controls of gold deposit formation. This has subsequently driven are-assessment of greenfields exploration targets. The current exploration concept has been proven to be effective, with both the discovery of Gounkoto and the successful replenishment of depleted Mineral Resources and Ore Reserves at both mines.

Mining and Ore Reserves

The open pit operations at Loulo-Gounkoto consists of multiple open pits. The open pits are operated by a mining contractor and adown-the-hole blasting service is provided by an

18^th September 2018 Page 21

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

appropriate blasting contractor. Opportunities exist to upgrade and convert Inferred Mineral Resource within the current pit shells to Ore Reserves with drilling. The open pit operations provide a source of flexibility to supplement the plant feed when required. The Gounkoto open pit mining will be completed in 2024 with the Faraba Pit starting in 2022 and ending in 2024.

The Loulo underground mines are mature operations designed to extract the Yalea and Gara orebodies. The Loulo underground mines use three primary mining methods consisting of long hole transverse open stoping; long hole longitudinal retreat open stoping; and stoping under rock fill (SURF). The Yalea LOM extends to 2028 and Gara to 2032. Yalea and Gara sustain production rates of approximately 1.45 Mtpa and Gara 1.25 Mtpa, respectively.

Gounkoto underground is atpre-feasibility level. The proposed mining method for Gounkoto underground consists of longhole bench stoping with backfill. Due to the relatively small reserve tonnage of Gounkoto underground, all attempts to reduce the capital outlay have been adopted. The MZ3 mineralisation located on the footwall of the Fault Gouge fault has been removed from the schedule due to the poor ground conditions modelled in the area. Construction of the Gounkoto underground operation is currently scheduled to start construction at the end of the open pit mine life in 2024, achieve first production in 2025, and continue until 2030. This is currently expected to be reviewed during 2018. The Gounkoto underground operation will commence in 2025 and will continue until 2030.

Randgold has significant experience in other mining operations within West Africa and the production rates, modification factors, and costs are benchmarked against other West African operations.

The current Ore Reserves for Loulo-Gounkoto support a total mine life of 15 years, mining a total of 52.2 Mt of ore at 4.73 g/t Au until the year 2032. During this15-year period the total plant ore feed tonnage will amount to 56 Mt, including current stockpiles, at an average feed grade of 4.55 g/t Au resulting in 7.5 Moz recovered at an average processing recovery of 92%.

The Qualified Person considers the parameters used in the Mineral Resource to Ore Reserve conversion process to be appropriate.

The Qualified Person is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, fiscal, infrastructure, permitting, that could materially affect the Ore Reserve estimate.

Processing

Based upon metallurgical testwork and actual operational data, the Qualified Person is satisfied that Loulo-Gounkoto is able to maintain production, gold recovery and reagent consumptions as forecasted.

The Loulo-Gounkoto has demonstrated successful operation both in terms of processing throughput and in particular with gold recovery. The average gold recovery in 2017 was 92.7%, an improvement over 2016.

18^th September 2018 Page 22

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gold recovery from Yalea ore is impacted by the presence of arsenic and copper. Arsenic and copper estimations are completed as part of the Mineral Resource update in order to identify potentially low recovery areas gold recovery.

The current LOM has an average recovery life of mine of 92%. Above 90% recovery is achieved by blending the various ore sources (Yalea / Gara/ Gounkoto) to control copper and arsenic grades in the mill feed.

The Qualified Person considers the modelled recoveries for all ore sources and the process plant and engineering unit costs applied to the Mineral Resource and Ore Reserve process to be acceptable.

Environment and Social

Loulo-Gounkoto has a mature environmental and social management plan and an accredited ISO14001 Environmental Management System in place which addresses current operational needs and can readily be adapted to meet future activities. Mine closure costs are reviewed and revised annually in line with good industry practice.

All permits are in place, and an annual Environmental and Social report compiled in a format aligned with GRI (Global Reporting Initiative) requirements is submitted to the Malian authorities.

Stakeholder engagement is ongoing, and all senior management are involved in regular meetings with the community. The mine prioritises local employment and in 2017 achieved 96% Malian employment across Randgold and contractor workforces.

Randgold continues to invest in community development initiatives, focussing on potable water supplies, primary school education, health care education, investment in medical clinics and local economic development projects, and livelihood projects, such as the program to improve the agricultural yield of the area. The mine is a significant employer to members of the local communities. The mining operations contribute to extendedlife-of-mine, employment of local Malians and the growth of the Malian economy. Randgold’s policy is to promote nationals to manage the Project.

The ongoing presence of artisanal miners (orpailleurs) operating within the Loulo Permit area, has the potential to cause unrest. The government of Mali provides security in the form of Gendarmes who intervene on request if ASM activities interfere with the operation of the mine. The risk of incursion into exploration or operations areas remains, because of the increasing number of people in the largely illegal activity. Dedicated orpaillage corridors have been proposed by the Mine but are still to be created for artisanal and small-scale mining. In the meantime, the Mine is reinforcing its relationships with the community to deal with the issue.

The Qualified Person considers the extent of all environmental liabilities, to which the property is subject, to have been appropriately met.

18^th September 2018 Page 23

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Ownership

All Permit fees, surface rights fees, and taxes relating to Loulo-Gounkoto exploration and mining rights have been paid to date, and reporting requirements conformed to, accordingly the concession is in good standing At the time of compiling this report, the Qualified Person is not aware of any risks that could result in the loss of ownership of the deposits or loss of the permits, in part or in whole.

Infrastructure

As a result of the long-established open pit mining operation at the Project, substantial infrastructure exists at Loulo-Gounkoto to support theon-going mining and processing operations.

The recently constructed Millennium Highway crosses the Loulo to Gounkoto haul road approximately 6 km north of Gounkoto Road providing access for staff and materials.

There is sufficient power supply capacity available from theon-site light and heavy fuel oil generators to meet the power demands of the operations.

An adequate water supply is available for the operation, sourced from the Gara and Falémé rivers which run through the Project site.

1.17 Risks

Randgold has undertaken analysis of the Project risks. Table1-8 and Table1-9 summarises the Project risks and the Qualified Person’s assessment of the risk degrees and consequences, as well as ongoing/required mitigation measures. The Qualified Persons, however, note that the degree of risk refers to our subjective assessment as to how the identified risk could affect the achievement of the Project objectives.

Loulo underground has been in production for 11 years and is a mature operation. Gounkoto open pit has been in production for over seven years and is a mature operation. Apre-feasibility study has been completed on Gounkoto Underground.

In the Qualified Persons opinion, there are no significant risks and uncertainties that could reasonably be expected to affect the reliability or confidence in the exploration information, Mineral Resource or Ore Reserve estimates.

Risk Analysis Definitions

The following definitions have been employed by the Qualified Persons in assigning risk factors to the various aspects and components of the Project:

● Low– Risks that are considered to be average or typical for a deposit of this nature and could have a relatively insignificant impact on the economics. These generally can be

18^th September 2018 Page 24

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

● Minor – Risks that have a measurable impact on the quality of the estimate but not sufficient to have a significant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

● Moderate – Risks that are considered to be average or typical for a deposit of this nature but could have a more significant impact on the economics. These risks are generally recognisable and, through good planning and technical practices, can be minimised so that the impact on the deposit or its economics is manageable.

● Major – Risks that have a definite, significant, and measurable impact on the economics. This may include basic errors or substandard quality in the basis of estimate studies or project definition. These risks can be mitigated through further study and expenditure that may be significant. Included in this category may be environmental/socialnon-compliance, particularly in regard to Equator Principles and IFC Performance Standards.

● High – Risks that are largely uncontrollable, unpredictable, unusual, or are considered not to be typical for a deposit of a particular type. Good technical practices and quality planning are no guarantee of successful exploitation. These risks can have a major impact on the economics of the deposit including significant disruption of schedule, significant cost increases, and degradation of physical performance. These risks cannot likely be mitigated through further study or expenditure.

In addition to assigning risk factors, the QPs provided opinion on the probability of the risk occurring during the LOM. The following definitions have been employed by the QPs in assigning probability of the risk occurring:

● Rare– The risk is very unlikely to occur during the Project life.

● Unlikely – The risk is more likely not to occur than occur during the Project life.

● Possible – There is an increased probability that the risk will occur during the Project life.

● Likely – The risk is likely to occur during the Project life.

● Almost Certain – The risk is expected to occur during the Project life.

18^th September 2018 Page 25

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table1-8 Loulo Risk Rating

Issue	Likelihood	Consequence Rating	Risk Rating	Mitigation
Geology andMineralResources – Confidence in Mineral Resource Models	Unlikely	Minor	Low	Additional scheduled infill drilling. Resource model updated on a regular basis using production reconciliation results.
Mining and Ore Reserves – Open Pit Slope Stability	Unlikely	Minor	Minor	Continuedin-pit monitoring, geotechnical drilling, instrumentation, and continued updating of geotechnical models.
Mining and Ore Reserves – Underground Recovery and Dilution	Possible	Moderate	Low	Extensive production drilling for stope design, including geotechnical features. Predictive modelling of dilution.
Processing – Process Water Management with Mine in Water-Positive Mode	Possible	Moderate	Moderate	Water balance investigations Processing to reduce high usage of fresh water I-plant water treatment has reduced deleterious elements
Processing - Baboto oxide effect on paste fill	Unlikely	Moderate	Low	Several test trials campaigns already successfully completed, which has identified the maximum content of Baboto ore that can accompany a plant feed. Plant trials have demonstrated that blending of oxides, provided that threshold contents are not exceeded, will not compromise the paste plant operation.
				Temporary holding of supernatant in redundant pits.
				Expansion of TSF.
Environmental – TSF Failure	Possible	Moderate	Moderate	Plant Water balance being addressed. Pool management and removal of excess water off the dam Continuing monitoring and external or third-party audits.
Environmental – Groundwater contamination (As)	Possible	Moderate	Moderate	Manage As levels of underground and TSF discharge through appropriate dilution techniques. Recycle contaminated water Continuing monitoring and external or third-party audits.

18^th September 2018 Page 26

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Issue	Likelihood	Consequence Rating	Risk Rating	Mitigation
Social – Social License to Operate	Possible	Moderate	Low	Regular community and union engagement by company social and sustainability representatives.
Social – Artisanal Miners	Likely	Moderate	Moderate	Engagement with Malian Government.to agree an ASM strategy. Gendarmes securing Permit area.
Country & Political – Security – Governmental	Possible	Major	Low	Dedicated government liaison team in Bamako. Government participation/ownership.
Capital and Operating Costs	Possible	Minor	Low	Continue to track actual costs and LOM forecast costs, including considerations for inflation and foreign exchange. Owner/Operator for underground mining.
Fiscal Stability	Possible	Moderate	Moderate	Dedicated government liaison team in Bamako. Government participation/ownership

18^th September 2018 Page 27

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table1-9 Gounkoto Risk Rating

Issue	Likelihood	Consequence Rating	Risk Rating	Mitigation
Geology and Mineral Resources – Confidence in Mineral Resource Models	Unlikely	Minor	Low	Additional scheduled infill drilling. Resource model updated on a regular basis using production reconciliation results.
Mining and Ore Reserves – Open Pit Slope Stability	Unlikely	Major	Moderate	Improved blast hole control. Continued in-pit monitoring.
Mining and Ore Reserves – Underground Recovery and Dilution	Possible	Moderate	Moderate	Further studies as part of the Feasibility Study.
Environmental – Air Quality (Dust)	Likely	Minor	Moderate	Dedicated dust suppression on haulage roads.
Social – Social License to Operate	Possible	Moderate	Low	Regular community and union engagement by company social and sustainability representatives.
Social – Artisanal Miners	Likely	Moderate	Moderate	Engagement with Malian Government.to agree an ASM strategy. Gendarmes securing Permit area.
Country & Political – Security – Governmental	Possible	Major	Low	Dedicated government liaison team in Bamako. Government participation/ownership.
Capital and Operating Costs	Possible	Minor	Low	Continue to track actual costs and LOM forecast costs, including considerations for inflation and foreign exchange.
Fiscal Stability	Possible	Moderate	Moderate	Dedicated government liaison team in Bamako. Government participation/ownership

18^th September 2018 Page 28

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

1.18 Recommendations

The Qualified Persons make the following recommendations:

● To extend the Loulo-Gounkoto life and replace depleted reserves, the current exploration strategy of targeting extensions of existing brownfields targets should be continued.

● A digital logging and data capture system should be implemented in the near future to minimise manual data capture.

● To reduce the complexity ofsub-domaining and improve the quality of local estimation, dynamic anisotropy should be tested and applied to folded bodies in future Mineral Resource updates.

● The application of a variablecut-off grade at Gara should be reviewed to avoid to the inclusion of large tonnages of lower grade material from the deeper levels of the mine impacting the profitability of the Gara operation.

● The Gara Dilution Rating System (DRS) model should be compared to production actuals, in order to establish if it could be directly applied to the Ore Reserve.

● To minimise the possibility of increased mining induced stresses affecting pillar and stope stability, a geotechnical review (including numerical modelling if appropriate) of the current Loulo underground mine plan and stoping sequence should be completed.

● The costs being used to calculate thecut-off grade for Gounkoto underground needs to be updated with the current underground costs and thecut-off grade applied to the next Mineral Resource and Ore Reserve estimates.

● It is recommended that a Qualified Person development programme be implemented at the Loulo-Gounkoto for resource geologists, mining engineers, and metallurgists.

● Loulo-Gounkoto is reliant on its own power generation through thermal energy sources. It is recommended that trials of alternative energy solutions continue to be sought as a potential reduction in power costs could have a material impact on the operating costs and mine life.

● An ASM strategy should be sought with the Malian government, and dedicated ASM corridors defined and managed.

18^th September 2018 Page 29

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

2 Introduction

2.1 Introduction

The purpose of this report is to support the public disclosure of the year end 2017 Mineral Resource and Ore Reserve estimates at the Loulo-Gounkoto Gold Mine Complex (Loulo-Gounkoto, the Mine or the Project), consisting of the Loulo Gold Mine (Loulo) and the Gounkoto Gold Mine (Gounkoto) located in western Mali. This Technical Report conforms to National Instrument43-101- Standards of Disclosure for Mineral Projects (NI43-101). All currency in this report is US dollars ($) unless otherwise noted.

Société des Mines de Loulo SA is an exploration and mining company and the owner of the Loulo mine, which is held 80% by Randgold Resources Limited, (Randgold) and 20% by the state of Mali.

Société des Mines de Gounkoto SA is an exploration and mining company and the owner of the Gounkoto mine which is held 80% by Randgold and 20% by the state of Mali.

Randgold is the operator of Loulo and Gounkoto. Randgold is developing, and operating gold mines in West and East Africa. The most notable of these are the following:

☐ Morila Gold Mine in Mali.

☐ Tongon Gold Mine in Côte d’Ivoire.

☐ Kibali Project in Democratic Republic of Congo (DRC).

☐ Massawa Exploration Project in Senegal.

The Project is an operating mine site comprising two underground mines (Yalea and Gara) at Loulo, the Gounkoto open pit mine, satellite deposits, and a processing plant (4.8 Mtpa capacity), together with other associated mine operation and regional exploration infrastructure. The plant produces gold doré bars.

Total mine production from underground and open pits in 2017 was 4.9 Mt at a head grade of 5.0 g/t Au for a total of 730 koz of gold (92.7% recovery).

Total production since mining commenced to year end 2017 is 391 Mt (53 Mt ore) for 5.6 Moz of gold (90.2% recovery).

2.2 Effective Date

The effective date of this report is 31^st December 2017.

18^th September 2018 Page 30

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

2.3 Sources of Information

This Technical Report was prepared by Randgold and incorporates the work of Roscoe Postle Associates Ltd (RPA) and Digby Wells and Associates Pty Ltd. (Digby Wells) (QPs). The dates of personal inspections of Loulo-Gounkoto by the QPs are provided in Section 29 of this Technical Report.

The QPs and their responsibilities for this Technical Report are listed in Section 29 Certificate of QP.

● Mr. Simon Bottoms, CGeol, MGeol, FGS, MAusIMM (Randgold), is responsible for the preparation of sections 2 to 12, 14, and 23 to 24 of this Technical Report and shares responsibility with hisco-authors for sections 1, 25, 26, and 27.

● Mr. Rodney B. Quick MSc, Pr. Sci.Nat (Randgold), is responsible for the preparation of sections 19, and 22 of this Technical Report and shares responsibility with hisco-authors for sections 1, 21, 25, 26, and 27.

● Mr. Richard Quarmby, BSc, Pr Eng & C Eng, MSAIChE, MIoMMM, MBA (Randgold), is responsible for the preparation of sections 13, 17, and 18 of this Technical Report and shares responsibility with hisco-authors for sections 1, 21, 25, 26, and 27.

● Mr. Derek Holm, BSc, FSAIMM, (RPA) is responsible for the preparation of sections 15 and 16 of this Technical Report, shares responsibility with hisco-authors for sections 1, 25, 26, and 27.

● Mr. Graham E. Trusler, Pr Eng, MIChE, MSAIChE (Digby Wells), is responsible for the preparation of section 20 of this Technical Report and shares responsibility with his coauthors for sections 1, 25, 26, and 27.

The documentation reviewed, and other sources of information, are listed at the end of this report in Section 27 References.

18^th September 2018 Page 31

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

2.4 List of Abbreviations

AARL	Anglo American Research Laboratory
ACACIA              Knelson Concentrates Circuit
AGC/AdvGC      Advance Grade Control
ALS	ALS Laboratories
AMTEL AMTEL Laboratory, Canada
ARD	Acid Rock Drainage
ASM	Artisanal and Small-Scale Mining
BHP	BHP Minerals
BM	Block Model
BRGM	Bureau de Recherches Géologiques et Minières
BRT	Bottle Roll Test
CA	Confidentiality Agreement
CAF	Cemented Aggregate Fill
CAPEX Capital Expenditure
CIL	Carbon in Leach
CIM	Canadian Institute of Mining, Metallurgy and Petroleum
CIP	Carbon in Pulp
CRF	Cemented Rock Fill
CRM	Certified Reference Material
CSTT	CSTT AO GROUP
CV	Coefficient of Variation
DCS	Distributed Control System
DD/DDH            Diamond Drill Hole
DGPS	Differential Global Positioning System
DL	Detection Limit
DO	Dissolved Oxygen
DRC	Democratic Republic of the Congo
DRS	Dilution Rating System
DTM	Digital Terrain Model
DTP	DTP Company, subsidiary of Bouygues
EDA	Estimation Data Analysis
EGL	Effective Grinding Length
EIA	Environmental Impact Assessment
EKATO Supplier of Agitators
EMP	Environmental or Emergency Management Plan
EMS	Environmental Management System

EOM	End of Month
EOY	End of Year
EPS	Datamine Enhanced Production Scheduler Software
ESIA	Environmental and Social Impact Assessment
EW	Electro-Winning
FGO	Full Grade Ore
FW	Footwall
FWFE	Footwall Far East
FWNE	Footwall North East
GC	Grade Control
GFSE	Gara Far South Extension
GM	General Manager
GPS	Global Positioning System
GRI	Global Reporting Initiative
HDPE	High Density Polyethylene
HFO	Heavy Fuel Oil
HPGR	High Pressure Grinding Rolls
HQ	Barrel Size (63.3 mm)
HV	High Voltage
HW	Hanging Wall
ICMC	International Cyanide Management
ID	Inverse Distance
IFC	International Finance Corporation
IMIU	International Mining Industry Underwriters
ITCZ	Inter Tropical Convergence Zone
JORC	Joint Ore Reserves Committee
KC	Knelson Concentrator
KE	Kriging Efficiency
LFO	Light Fuel Oil
LHOS	Long Hole Stope Mining
LIDAR	Light Detection and Ranging
LIMS	Laboratory Information Management System
LOM	Life of Mine
MASL	Metres Above (Mean) Sea Level
MBA	Master of Business Administration

18^th September 2018 Page 32

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

MCC	Motor Control Centre
MCF	Mine Call Factor
MMD	MMD Group of Companies
MO	Marginal Ore
MRM	Mineral Resource Management
MRMM	Mining Rock Mass Model
MSO	Minable Stope Optimiser
MVA	Mega Volt Amp
MW	Megawatt
MZ	Main Zone
NPV	Net Present Value
NQ	Core Size (47.6 mm)
NSR	Net Smelter Royalty
OC	Open Cast
ODBC	Open Database Connectivity
OEM	Original Equipment Supplier
OHSAS Occupational Health and Safety Assessment Series
OK	Ordinary Kriging
OMC	Orway Mineral Consultants
OP	Open Pit
OPEX	Operating Costs
OREAS ORE Research & Exploration Pty Ltd CRM Manufacture
PFS	Prefeasibility Study
PP	Purple Patch
PQ	Core Size (85.0 mm)
PSA	Pressure Swing Adsorption
QA/QC	Quality Assurance/Quality Control
QG	QG Australia Ltd
QKNA/KNA        Quantitative Kriging Neighbourhood Analysis
QP	Qualified Person
QQ	Quantile-Quantile
QT	Quartz-Tourmaline Unit
RAB	Rotary Air Blasted
RAP	Resettlement Action Plan

RC	Reverse Circulation
RES	Resource Domain
RL	Elevation (m)
RMR/MRMR      Rock Mass Rating (Mean)
ROM	Run of Mine
SAG	Semi-Autogenous
SAP	Saprolite or German Company
SCADA Supervisory Control And Data Acquisition
SCN	Sodium Cyanide
SD	Standard Deviation
SENET Engineering Company
SG	Specific Gravity
SGS	SGS Laboratories
SLC	Sub Level Caving
SLTO	Social License to Operate
SOMILO            Société des Mines de Loulo
SOP	Standard Operating Procedure
SQL	Structured Query Language Database
SQR	Argillaceous Quartzites
SR	Slope of Regression
SRK	Steffen Roberts and Kirsten, Engineering Company
SURF	Stoping Under Rock Fill
SWZ	Southwest Zone
SZ	Southern Zone
TDS	Total Dissolved Solids
TL	Tolerance Limit
TSF	Tailings Storage Facility
UG	Underground
UTM	Universal Transverse Mercator
VHF	Very High Frequency
VOC	Volatile Organic Compounds
VSAT	VSAT Internet Connection
WAD	Weak Acid Dissociated
WRD	Waste Rock Dump
ZADRA Type of elution circuit

18^th September 2018 Page 33

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

2.5 Units

cm	Centimetre
ekW	Generator Output Rating in kW
g	Grammes
Ga	Billion years
g/cm3	Grammes per Cubic Centimetre
g/t	Grammes per Metric Tonne
Ha	Hectare
Kbar	Kilobar of pressure
kg	Kilogram
km	Kilometre
km2	Square kilometre
koz	Thousand ounces
kt	Thousand metric tonnes
ktpa	Thousand metric tonnes per annum
ktpm	Thousand tonnes per month
kW	Kilo Watts
m	Metre
m²	Square meter
m3	Cubic meter
Mm3	Million Cubic Metres
Ml	Million litres
Moz	Million fine troy Ounces
Mt	Million Metric tonnes

Mtpa			Million tonnes per annum
MVA			Mega Volt Amps
MW			Mega Watts
oz			Fine troy ounce equalling 31.10348
			grams
ppm			Parts per million
t			Metric tonne
tm-3			Density measured as metric tonnes
			per cubic metre
°			Degrees
	‘		Minutes
%			Percentage
%w/v			Percentage Weight by Volume
µm			Microns
Ø			Diameter
#			Mesh
$			United States Dollar
$		‘000	Thousand United States Dollars
$		M	Million United States Dollars
$/oz United States Dollar per ounce
$/t United States Dollars per Metric Tonne

18^th September 2018 Page 34

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

3 Reliance on Other Experts

This report has been prepared by Randgold. For the purpose of this report, the QPs have relied upon information provided by Randgold’s Legal Counsel regarding the validity of exploitation permits; this opinion has been relied upon on in Section 4 (Property Description and Location) and in the summary of this report.

18^th September 2018 Page 35

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

4 Property Description and Location

4.1 Project Location

Loulo-Gounkoto is situated in Western Mali adjacent to the Falémé River which forms the international boundary with Senegal. The Republic of Mali is a landlocked country and is bordered by Guinea, Senegal, Mauritania, Algeria, Niger, Burkina Faso, and the Cote d’Ivoire.

The Project area is located 350 km west of the capital city of Bamako, 220 km south of the town of Kayes, and NW of the nearest town Kenieba (Figure4-1). It falls within the Central Arrondissement of the Kenieba District which is one of the 10 districts of the Kayes Region.

18^th September 2018 Page 36

18th September 2018 Page 37

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

4.2 Mineral Rights and Land Ownership

The Loulo mine is within the Loulo Exploitation Permit (Loulo Permit). The original Loulo Permit was granted by Decret No.96-048/PM-RM 9, on the 14^th February 1996 and covered an area of 48 km². This Permit was amended by Decret No.99-193/PM-RM dated 15^th July 1999 and extended the size of the Loulo Permit to 372 km². The Loulo Permit was amended by DecretNo. 2012-311/P-RM on 21^st June 2012 which reduced the size of the Lolo Permit as the portion surrounding the Gounkoto Mine was transferred to a new Exploitation Permit. The Loulo Permit remains in force for a period of 30 years after which is renewable if production is still taking place.

In 2010 Randgold applied and was granted the formation of the new Gounkoto Exploitation Permit (Gounkoto Permit), which was split from the Loulo Permit, to form a separate entity, Société des Mines de Gounkoto SA (Gounkoto) under DecretNo.2012-431/PM-RM DU dated 3^rd August 2012. The Gounkoto Permit incorporates the Gounkoto and Faraba Reserves and is valid for 30 years.

In 2017, the Baboto North deposit was purchased by Endeavour Mining Corporation. This resulted in a minor change to the Loulo Permit. An updated Decret number has not yet been generated by the Malian Government, but it should be created imminently. Figure4-2 illustrates the current Loulo and Gounkoto Permit boundaries. Table4-1 details the coordinates of the current Loulo and Gounkoto Permits. The Loulo and Gounkoto Permits areas are currently 261.23 km² and 99.95 km² respectively, for a total area of 361.18 km².

18^th September 2018 Page 38

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table4-1 Loulo and Gounkoto Permit Coordinates

Permit	Point	Longitude	Latitude	Permit	Point	Longitude	Latitude
	A	11° 19’ 00”	13° 10’ 00”	Loulo	AF	11° 26’ 00”	13° 10’ 00”
	B	11° 19’ 00”	12° 56’ 33”		AG	11° 22’ 57”	13° 10’ 00”
	C	11° 20’ 50”	12° 56’ 33”		AH	11° 22’ 57”	13° 09’ 51”
	D	11° 20’ 50”	12° 57’ 24”		AI	11° 21’ 59”	13° 09’ 51”
	E	11° 21’ 25”	12° 57’ 24”		AJ	11° 21’ 59”	13° 09’ 41”
	F	11° 21’ 25”	12° 59’ 37”		AK	11° 21’ 00”	13° 09’ 41”
	G	11° 22’ 59”	12° 59’ 37”		AL	11° 21’ 00”	13° 10’ 00”
	H	11° 22’ 59”	12° 58’ 06”	Gounkoto	A	11° 19’ 00”	12° 56’ 33”
	I	11° 23’ 37”	12° 58’ 06”		B	11° 19’ 00”	12° 50’ 00”
	J	11° 23’ 37”	12° 58’ 28”		C	11° 23’ 38”	12° 50’ 00”
	K	11° 24’ 26”	12° 58’ 28”		D	11° 23’ 38”	12° 49’ 39”
	L	11° 24’ 26”	12° 59’ 26”		E	11° 24’ 11”	12° 49’ 39”
	M	11° 24’ 49”	12° 59’ 26”		F	11° 24’ 11”	12° 50’ 39”
	N	11° 24’ 49”	13° 00’ 08”		G	11° 24’ 32”	12° 50’ 39”
	O	11° 23’ 44”	13° 00’ 08”		H	11° 24’ 32”	12° 51’ 00”
Loulo	P	11° 23’ 44”	13° 01’ 07”		I	11° 23’ 45”	12° 51’ 00”
	Q	11° 24’ 35”	13° 01’ 07”		J	11° 23’ 45”	12° 52’ 09”
	R	11° 24’ 35”	13° 02’ 25”		K	11° 24’ 03”	12° 52’ 09”
	S	11° 25’ 11”	13° 02’ 25”		L	11° 24’ 03”	12° 53’ 30”
	T	11° 25’ 11”	13° 03’ 14”		M	11° 23’ 43”	12° 53’ 30”
	U	11° 25’ 54”	13° 03’ 14”		N	11° 23’ 43”	12° 53’ 50”
	V	11° 25’ 54”	13° 04’ 00”		O	11° 23’ 14”	12° 53’ 50”
	W	11° 25’ 29”	13° 04’ 00”		P	11° 23’ 14”	12° 54’ 42”
	X	11° 25’ 29”	13° 05’ 23”		Q	11° 23’ 59”	12° 54’ 42”
	Y	11° 26’ 08”	13° 05’ 23”		R	11° 23’ 59”	12° 55’ 43”
	Z	11° 26’ 08”	13° 05’ 52”		S	11° 23’ 33”	12° 55’ 43”
	AA	11° 28’ 28”	13° 05’ 52”		T	11° 23’ 33”	12° 55’ 27”
	AB	11° 28’ 28”	13° 06’ 13”		U	11° 22’ 26”	12° 55’ 27”
	AC	11° 28’ 42”	13° 06’ 13”		V	11° 22’ 26”	12° 55’ 45”
	AD	11° 28’ 42”	13° 07’ 00”		W	11° 20’ 50”	12° 55’ 45”
	AE	11° 26’ 00”	13° 07’ 00”		X	11° 20’ 50”	12° 56’ 33”

18^th September 2018 Page 39

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure4-2 Loulo and Gounkoto Permit Areas

4.3 Surface Rights

The holder of the Permit may freely carry out exploration and mining activities within this area provided that there is compliance with government regulations including those of safety and the environment. The Permit allows the holder to fully utilise the surface for the implementation of such infrastructure as required for the mining operation. All the taxes relating to Loulo and Gounkoto Mining Rights have been paid to date and the concession is in good standing. There are no exclusion zones on the Permit.

The QP is not aware of any other significant factors and risks that may affect access, title, or the right of ability to perform work on the property.

18^th September 2018 Page 40

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

4.4 Ownership, Royalties and Lease Obligations

The Loulo Permit is owned by Société des Mines de Loulo SA which is held 80% by Randgold and 20% by the state of Mali.

The Gounkoto Permit is owned by Société des Mines de Gounkoto SA which is held 80% by Randgold and 20% by the state of Mali.

The Loulo-Gounkoto Establishment Convention regulates the fiscal conditions under which the Loulo and Gounkoto mine operates and is based on the1991 Mining Code. A 6% royalty is payable to the Malian government based upon production together with a corporate tax rate on profits at 30% and a minimum of 0.75% on gross revenues if a loss is made. Loulo received a five year tax holiday from first commercial gold production in October 2005. Gounkoto received a two year tax holiday from first gold production in 2013 and has since received governmental approval for use of 50% corporate tax reduction for the next four years to support its development of a super pit. The convention includes exoneration on fuel duties for the life of the project and on import duties for three years from the start of first gold.

Other than the ownership status, royalty on the revenues from mineral production, there are no other royalties,back-in rights, payments, or other agreements and encumbrances to which the property is subject.

There are no known environmental, permitting, legal, title, fiscal, socio-economic, marketing, and political or other issues known at this time that could materially affect the Mineral Resource declared.

The QP is not aware of any risks that could result in the loss of ownership of the deposits or loss of the permits, in part or in whole.

18^th September 2018 Page 41

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

5 Accessibility, Climate, Local Resources, Infrastructure and Physiography

5.1 Accessibility

Primary access for staff and consumables to Loulo-Gounkoto is via the recently constructed (2011) Millennium highway that runs from Dakar, Senegal to Bamako, Mali. The Millennium Highway crosses the Loulo to Gounkoto haul road approximately 6 km north of Gounkoto and provides excellent road connections in comparison to much of the country.

Daily flights with international air carriers are available from Dakar and Bamako. Charter flights between Bamako and the (unsealed) airstrip at the mine are regularly used. The landing strip at Loulo is approximately 1.5 km long and built from laterite material. It can accommodate moderate sized aircraft and has been awarded full certification by the transport authorities in Mali.

5.2 Climate

The climate at Loulo-Gounkoto is strongly influenced by the north and southward movement of the Inter Tropical Convergence Zone (ITCZ) which creates distinctive wet and dry seasons. The site is in the Sahelian Transition Zone between the Sahara Desert in the north and the tropical climate in the south. The low altitude of the site (90 m to 120 m above mean sea level) and the absence of any intervening mountains mean that the humidity is directly conveyed to the site when the wind blows in that direction.

summarises the monthly rainfall as measured at the Gounkoto and the Loulo stations. Using the average values for all the years on record, it shows a strongly unimodal rainfall distribution, with 87% of rain falling between the months of June and September. The mean annual precipitation amounts to 1,091 mm.

Based on regional data (Kenieba), the potential evaporation is estimated to range between 105 mm and 200 mm per month. In total, this amounts to an annual potential evaporation of 1,749 mm. Although the annual evaporation thereby exceeds the annual rainfall, an excess of water is available during the peak of the wet season (July to September).

The site is situated within close proximity of the Sahelian Transition Zone between the Sahara Desert in the north and the tropical climate in the south. Temperatures range between approximately 13°C and 43°C (average 28°C) with the hottest months between March and June.

Climatic conditions do not materially affect either exploration, development, or mining operations.

18^th September 2018 Page 42

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table5-1 Monthly Records of Precipitation and Potential Evaporation

Month	Jan		Feb		Mar		Apr		May		Jun		Jul		Aug		Sep		Oct		Nov		Dec		Total
Statistic	(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)		(mm)
Average Rainfall		2		3		6		4		42		142		217		325		263		69		14		6		1,091
Kenieba - Average Potential Evaporation		167		174		200		188		83		142		119		105		111		120		110		130		1,749

5.3 Physiography

Topography of the Gounkoto area is generally flat with elevations ranging from 100 m above sea level to a maximum of 200 m above sea level. Laterite or iron cap development is a common feature throughout the area. The Project lies in a low seismic rated area.

Within the Project area the landscape is characterised by scrubland and brush vegetation.

5.4 Local Resources

The local population are essentially subsistence farmers who supplement their income through artisanal gold mining. Bread and small quantities of vegetables may be sourced locally but most supplies are obtained from Kayes or Bamako. As part of Randgold’s strategy through the provision of correct training it has been possible to employ local people who make up almost all of the labour requirement at the Project.

Local infrastructure around Gounkoto is limited to small rural settlements connected by gravel roads and paths. Social and economic baseline studies identified that the inhabitants within the villages are mostly Malinké but include other ethnic groups who have arrived later such as Peul, Bambara, and Bozo. The main economic activities are agriculture and artisanal mining. Other activities include livestock, fishing, and trade. All the houses are traditional round huts covered with thatch. Forty-seven boreholes and four water supply systems have been constructed in the host communities in total, 16 schools have been built which has increased student enrolment to more than 4,764 from a base of 500 prior to the mine being built. Randgold has a target of a school in each village which has been achieved in Loulo.

Local food chains have been improved through the donation and the building of infrastructure such as dams and roads along with the launch and support of an Agribusiness hub in the region. A number of local entrepreneurs have utilised these such opportunities and increased economic activity in the region.

Health facilities have been improved with a health centre being built and two other centres being rehabilitated. A partnership with anon-government organisation is in place which targets the spread of human immunodeficiency virus (HIV). Malaria infections have also been targeted in local communities.

Randgold’s policy promotes local employment. Figures for national employment in 2017 were 2,820 out of a total of 2,975 (95%) employee and contractor posts at Loulo and 1,175 out of 1,209

18^th September 2018 Page 43

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

(97%) employee and contractor posts at Gounkoto. Unskilled labour is typically sourced from the local area while more skilled posts are filled by staff from elsewhere in Mali, including Bamako.

5.5 Infrastructure

The supply route for Loulo-Gounkoto runs from the port of Dakar in Senegal. The border with Senegal is within 3 km of the mine and the road crossing for goods is approximately 40 km along an all weather laterite surfaced road to Loulo and 10 km to Gounkoto. Local infrastructure is limited to small rural settlements connected by gravel roads and paths. The Dakar to Bamako Millennium highway runs approximately 6 km north of the Project.

The Loulo mine is an operating mine site comprising open pit and underground mines, a processing plant, satellite deposits and associated infrastructure. Previously mined open pits remain open and are used to access the underground mine. Waste dumps are situated adjacent to the open pits. The plant and offices and accommodation village are located east of the Gara pit. The TSF is located 8 km to the east of the plant. The Loulo and Gounkoto mine site layouts are shown in Figure5-1 and Figure5-2.

Water is sourced for the Project from the Gara and Falémé rivers which run through the Project site and from water generated underground. A monthly report is generated to monitor this.

Power is generated on site using light and heavy fuel generators. Substantial infrastructure has been built at Loulo-Gounkoto as a result of the long-established open pit mining operation. This includes ore processing and tailings facilities, workshops offices, and camps. Mobile telephone services are available across the mine.

18^th September 2018 Page 44

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18^th September 2018 Page 45

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18^th September 2018 Page 46

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

6 History

6.1 Historical Exploration and Development

Gold potential in the Loulo area was recognised by the Syndicat Or joint venture between the Malian Direction Nationale de la Géologie et des Mines and the French Bureau de Recherches Géologiques et Minières (BRGM). As a result of exploration work undertaken by the Syndicat Or joint venture, the Loulo 0 gold deposit (more recently renamed to Gara) was discovered in 1981. Syndicat Or continued with exploration until 1989, concluding with apre-feasibility study indicating that the Loulo 0 deposit (Gara) on its own wassub-economic.

In 1992, after an evaluation of the available data, BHP Minerals Mali (BHP) entered into an Option Share purchase, and work commitment agreement with SOMILO (Société des Mines de Loulo) to purchase a 20% share of the joint venture wholly-owned by the Republic of Mali and SEREM (a 100% subsidiary of BRGM). This was later increased to 51% on the completion of a feasibility study.

BHP completed the following work:

● 20,158 m of core drilling, including 16,085 m at Loulo 0 (Gara).

● Two regional soil sampling programmes covering the mining and exploration concessions.

● Metallurgical testwork on the Loulo 0 core and additional bulk samples.

● Geotechnical studies.

● Preliminary open pit and underground mine designs.

Work focused on Loulo 0 (Gara) and BHP produced separate resource estimates for mineralisation that is potentially mineable from surface, mineralisation that must be mined underground, and mineralisation from satellite deposits.

As a result of this work BHP delineated an open pit Indicated Mineral Resource of 3.48 Mt at an insitu grade of 4.42 g/t Au, an underground Indicated Mineral Resource of 0.87 Mt at a grade of 9.7 g/t Au and an Inferred Mineral Resource of 1.00 Mt at a grade of 10 g/t Au. Reconnaissance drilling identified other Inferred Mineral Resources in outlying satellite deposits of 1.42 Mt at 4.6 g/t Au.

The conclusion from the feasibility work carried out by BHP was that the Loulo 0 (Gara) deposit by itself was too small to be economic and that an additional satellite deposit of at least 500 koz was required for an economically viable project. By spending $4.62 M on this work, BHP increased its share in Somilo to 20%.

Following the discovery of the Yalea deposit a series of feasibility studies were undertaken.

18^th September 2018 Page 47

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

6.2 Randgold Project Milestones and Development

A history of the key dates in Randgold’s involvement in Loulo-Gounkoto is presented below:

1996

● Randgold acquired BHP Minerals Mali in October 1996 and changed its name to Randgold Resources Mali.

1997

● Randgold increased its share of SOMILO to 51% in October 1997. At this stage, as a result of a corporate agreement, La Source (a joint venture between the BRGM and Normandy Limited) replaced BRGM in SOMILO.

● Discovery of Yalea deposit

● Listing of Randgold Resources on London Stock Exchange and initial public offering (IPO) raises $83 M used to fund Morila feasibility study and increase share in SOMILO to 51%.

2001

● Randgold acquired La Source’s 29% holding in SOMILO in April 2001.

2003

● An updated Loulo feasibility study was generated with total reserves of 1.38 Moz.

2004

● Construction begins on Loulo Mine.

2005

● Loulo mine opens with Yalea and Gara open pit mining.

● Loulo underground feasibility study approved.

2006

● Development of Yalea underground starts.

2008

● Yalea underground produces first gold.

2009

● Discovery of Gounkoto deposit.

2010

● Gara underground mine commences.

18^th September 2018 Page 48

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

2011

● First gold produced from Gounkoto open pit.

● First gold produced from Gara underground.

2012

● Loulo processing plant ramped up to over 4 Mt per annum.

● Mining of ‘Purple Patch’ at Yalea started.

● ISO 14001 achieved at Gounkoto.

● OHSAS 18001 safety accreditation achieved at Loulo.

2013

● Gara underground conveyor installed.

● Additional four CIL tanks installed and commissioned.

● Grinding and crushing upgraded to eliminate scats.

● Successful conversion of generators to Heavy Fuel Oil.

● MZ4 lode discovered at Gounkoto.

2014

● Gounkoto underground Ore Reserve defined.

● Production exceeded 600 koz gold.

● Loulo paste backfill plant installed.

● Primary vent shafts at Yalea and Gara installed.

● Secondary crushing plant upgraded.

2015

● Loulo underground transition to owner operator mining.

● Carbon regeneration kiln installed.

2016

● Production totalled 707 koz gold.

● Gounkoto ‘Super Pit’ slope designs completed.

● Gounkoto ‘Super Pit’ versus undergroundtrade-off study completed.

● Gounkoto undergroundpre-feasibility study published.

● Upgraded elution circuit.

● Completed underground refrigeration project.

2017

● Production increased to 730 koz gold.

18^th September 2018 Page 49

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

● Increased plant throughout to over 4.9 Mt.

● Increased plant recoveries to 92%.

● Commissioned new Yalea underground crusher.

● Started Gounkoto ‘Super Pit’ stripping.

● Sale of Baboto North to Endeavour Mining Corporation.

6.3 Historical Resource and Reserve Estimates

These estimates are considered to be historical in nature and should not be relied upon. A QP has not completed sufficient work to classify the historical estimate as a current Mineral Resource or Ore Reserve and Randgold is not treating the historical estimates as current Mineral Resources or Ore Reserves. These have been superseded by the Mineral Resource estimates reported and described in this report.

Loulo

BHP Minerals Mali delineated a Mineral Resource on the Loulo 0 (Gara) deposit prior to Randgold’s acquisition. The Mineral Resource estimate contained 3.5 Mt at 4.43 g/t Au of Open Pit Indicated material, 0.87 Mt at 9.7 g/t Au of underground Indicated material and 1.0 Mt at 10.0 g/t Au of Inferred material. An additional 1.4 Mt at 4.6 g/t Au of Inferred material was identified at outlying satellite deposits.

Gounkoto

Gounkoto was a greenfields discovery by Randgold and therefore there are no historical Mineral Resource estimates available.

6.4 Resource and Reserve Evolution

Figure6-1 and Figure6-2 present the evolution of cumulative Mineral Resources and Ore Reserves from the initial acquisition of Loulo and Randgold’s discovery of Gounkoto.

18^th September 2018 Page 50

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure6-1 Loulo Mineral Resource and Ore Reserve Evolution

Figure6-2 Gounkoto Mineral Resource and Ore Reserve Evolution

18^th September 2018 Page 51

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

6.5 Past Production

Since commencing mining operations in 2005 and the end of 2017, a total of 391 Mt (53 Mt ore) have been mined from the various orebodies at Loulo-Gounkoto.

Table6-1 details the past production from Loulo-Gounkoto since 2005.

Table6-1 Past Production Records for Loulo-Gounkoto

Year	Tonnes Milled (kt)	Head Grade (g/t)	Gold Produced (oz)	Recovery (%)
2005	551	4.1	67,984	94.3
2006	2,600	3.2	241,575	93.9
2007	2,700	3.3	264,647	93.2
2008	2,720	3.2	258,095	91.2
2009	2,947	4.2	351,591	87.7
2010	3,158	3.4	316,539	92.5
2011	3,619	3.4	346,179	88.1
2012	4,354	4.0	503,224	89.2
2013	4,463	4.6	580,364	88.4
2014	4,396	5.0	639,219	90.2
2015	4,543	4.8	630,167	90.1
2016	4,875	5.0	707,116	91.0
2017	4,918	5.0	730,372	92.7
Total	45,598	4.2	5,637,072	90.2

18^th September 2018 Page 52

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

7 Geological Setting and Mineralisation

7.1 Regional Geology

The West African Craton can be divided into three main regions exposed beneath Phanerozoic cover. In the north, the Reguibat Rise extends over Mauritania; in the west Algeria and consists of an Achaean terrain and the Proterozoic Birimian terrain is situated in the east. The southern Leo Rise covers a large area over southern Mali, Côte d’Ivoire, Burkina Faso, Niger, Ghana, and Guinea; and is separated from the Reguibat Rise by the Late Proterozoic to Phanerozoic sedimentary Taoudeni Basin. The western Achaean portion known as the Man Shield is separated from the eastern Birimian formation of the Baoule Mossi domain by the Sassandra fault.

Loulo-Gounkoto is located within the Kedougou-Kenieba erosional inlier (Figure7-1) which is underlain by Lower Proterozoic (2.1 Ga) Birimian meta-sedimentary-volcanic sequences.

Figure7-1 Location of the Kedougou-Kenieba-Inlier in the West African Craton

Volcanic and sedimentary rocks of the Birimian Supergroup form a substantial portion of the West African Craton. These rocks and associated intrusives of the Eburnean tectonic-thermal event represent a major Paleoproterozoic, juvenile crust forming event that took place during the time interval 2.25 Ga to 2.09 Ga. It is believed that the Birimian crust formed through the closure of an oceanic basin between two Achaean Cratons, involving the progressive accretion of island arcs and oceanic plateau to a growing continental mass. Subsequent thickening associated with the intrusion of kinetically late basin-type granitoids and compressional deformation, facilitated the formation of structurally controlled mesothermal type lode gold deposits.

18^th September 2018 Page 53

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

7.2 Local Geology

The Permits are located within the Kedougou-Kenieba erosional inlier. The inlier is unconformably overlain by Upper Proterozoic sandstones towards the east and further south. The Kenieba inlier contains several significant gold deposits including Sadiola, Yalea, Segala, Tabakoto, and Gounkoto deposits in Mali, and Sabodala in Senegal (Figure7-2). The regional setting of each of these deposits is characterised by its close proximity to a major structure

The area is extensively laterised and covered by regolith material, with only about 6% outcrop. Three formations are present in the region including:

● Kofi (shelf to deep marine sediments): consisting of greywacke, sandstone, argillaceous sandstone, calcareous sandstone, limestone, and tourmalinised sandstone units.

● Saboussire (andesitic volcanic arc with shelf sedimentary setting): andesite, basalt, volcanic breccia, and tuff.

● Keniebandi volcanic and sedimentary units.

The Senegal-Mali shear marks a major break in the geology from shelf carbonates with the Falémé ironstone unit in the west to the sedimentary sequences of the Kofi formation in the east.

The aero magnetics survey has highlighted the major tectonic fabrics, the Moussala granite, Falémé ironstone, the dolerite dyke, and a major circular feature. This circular structure is coincident with a zone of intense structure deformation (WNW and NE fabrics) and multiple quartz tourmaline units. The Gara and Yalea deposits and a large number of additional gold targets are located within this circular feature.

7.3 Permit Geology

The Loulo and Gounkoto permits are predominately underlain by the Kofi formation consisting of greywacke, sandstone, argillaceous sandstone, calcareous sandstone and tourmalinised sandstone units. The lithologies present are interpreted to represent a forearc environment with a sequence of older deep-water argillite and greywacke to the east of the permit area, and limestone and carbonated clastic sediments towards the west, representing a shallow shelf setting. This broad classification is complicated by a series of thrusts which have caused stacking and repetition of the sedimentary package. Regional structures which transgress lower order structures, rheologically contrasting lithologies and intrusive bodies have a strong spatial relationship with the occurrence of gold deposits. These regional structures strike for

18^th September 2018 Page 54

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18^th September 2018 Page 55

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

over 50 km north to NE across the permit area. The Permit area is transgressed by several regional linear structures which include:

● 1^st order Senegal – Mali Shear Zone.

● 2^nd order north to NNE trending Farandi and Yalea structures. The first mineralising phase was introduced during this event.

● 3^rd order NE to ENE trending reverse faults and north trendingre-activated shear zones. These cross and reactivated structures are present at all the major deposits and mineralised sites and have been responsible for focusing gold mineralisation.

Work to date indicates an overall compressional tectonic regime with oblique compression on the NE structures and reactivation of older NNE structures. Figure7-3 illustrates the position of the Loulo and Gounkoto targets in relation to the other targets and deposits on the Loulo and Gounkoto Permits.

7.4 Loulo Permit Geology and Mineralisation

The Loulo project area is characterised by a sequence of sandstone, limestones and greywackes which are truncated by three major north to NE trending shear zones. Two major mineralised bodies have been discovered within the Project area, Gara and Yalea. Several other satellite deposits are also present.

Yalea

Yalea is located in a north-south striking, steeply dipping package of metasedimentary rocks. Host lithologies at Yalea (from west to east) comprise of quartzite/grey in the hanging wall, with tectonic breccias in the north. Immediately above the main body of mineralisation is thin (0 m to 5 m) sequence of banded schistose greyish limestone, with alternating white and grey calcitic layers and dark grey to black phyllite units. The main mineralised body is a hydrothermally brecciated argillaceous pink quartzite that becomes more argillaceous (and less altered) towards the footwall. A higher grade ‘Purple Patch’ zone is observed in a dilatational strain transfer zone formed as the dip of the mineralised package steepens forming hydraulic breccias. The footwall package is a thick sequence of argillaceous quartzite and black sandstone. This sedimentary package is intruded locally by thin (0.1 m to 2 m) acid intrusives of mostly granitic composition. The country rocks are also cross cut by a late EW dolerite dyke that is generallysub-horizontal, with a gentle southward plunge (Figure7-4).

18^th September 2018 Page 56

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18^th September 2018 Page 57

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure7-4 Yalea Simplified Geology Cross Section

Alteration

Alteration in the Yalea deposit is characterised by intense carbonate alteration, albitisation, silicification, sericitisation and chloritisation. Alteration is invariably associated with sulphide but also forms a halo within the footwall and the hanging wall units. The alteration can be identified by the localisation of coarse-grained aggregates of carbonate, quartz, and albite, whereas the mineralisation is associated with chloritisation-sericitisation and ferrous carbonate.

Figure7-5 Early pink Albite ± Carbonate Alteration. B. Early Light Brown Albite±-Carbonate Alteration

and Narrow Bands of Sericite ±Chlorite

18^th September 2018 Page 58

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Weathering at Yalea extends up to 300 m below surface due to the strong carbonate alteration associated with the mineralised structure. Four main weathering types have been characterised in the rock, saprolite, transition, fresh, and high-grade fresh.

Structure

At Yalea, the main mineralised body is a hosted by the north-striking Yalea Shear, where it is intercepted by the Yalea Structure. The Yalea Shear is a brittle-ductile, north-south striking, mineralised fault that transects the Yalea Structure, which is a complex, of north to NNE striking shear zones.

Both major structures include variety of deformed and altered rocks, even where they are unmineralised. Much of the deformation, brecciation and barren alteration along both structures may have occurred during the regional-scale shortening and upending of the Birimian rocks prior wrench-style deformation associated with mineralisation. Both structures appear to have been a focus for hydrothermal activity over an extended period, culminating in the episode of hydrothermal activity associated with gold mineralisation. Gold mineralisation is situated between a hanging wall and footwall shear within the reactivated 2 km portion of the predominant Yalea – Baboto structure. The mineralisation dips to the west at 65° for the initial 200 m to 300 m vertically after which the structure steepens and becomes sub vertical.

Mineralisation

Economic levels of gold mineralisation are almost exclusively associated with paragenetically late sulphide veins, breccias and zones of massive sulphides. Higher grade material commonly contains sulphide veins which cut the various generations of albite ± carbonate alteration. There is a strong correlation between sulphide intensity and gold mineralisation with the dominant sulphide phases consisting of pyrite (abundant), arsenopyrite and minor chalcopyrite. The sulphides can be disseminated or massive along fabric.

Yalea mineralisation, remains open at depth and to the south with potential for significant high-grade extensions. To the south of the ‘Purple Patch’ zone the Yalea Shear forms asub-horizontally plunging transfer zone where the competency contrast of the footwall argillaceous quartzites (SQR) contact causes a transfer of strain associated with normal movement. Consequently, this transfer creates dilatational hydraulic breccias with the potential to host significant high-grade extensions (Figure7-6).

18^th September 2018 Page 59

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure7-6 Long Section Looking West Showing Structural Model for Yalea Deposit

Gara

Gara is located 6 km NNW of the Yalea deposit and is hosted within 800 m long tourmaline sandstone/greywacke unit which outcrops on surface as black quartzite forming small (10 m) topographic highs. The host lithologies are (from west to east): chemical (limestone or carbonate altered) sediments and alternating argillitic and quarzitic bands (argillaceous quartzite or SQR) in the hanging wall, mineralised quartz tourmaline (QT) ranging from 5 m to 20 m thick (average 15 m) and a coarse to medium grained greywacke unit in the footwall. The sedimentary package is also cross cut by three unmineralised late EW doleritesub-horizontal dykes that plunge shallowly from north to south.

Structure

The geometry of the deposit is dominated by the strike slip shearing on Senegal-Mali shear and the development of associated conjugate sets of antithetic structures. This shearing has resulted in folding, fracturing, brecciation, and development of a quartz vein stockworks within the QT which behaved in a more brittle manner during deformation.

The deposit as a whole, spans the hinge of a broad open fold with a gently-plunging north-south trending axis. The upper limb of this fold dips moderately to steeply west, whereas the lower limb dips moderately to steeply east. A second generation of younger smaller-scale southwest plunging, distinctlynon-cylindrical folds are developed in the upper part of the deposit. The distribution of gold grade in long section reflects the varying degrees of fracturing, brecciation, and subsequent development of a quartz-carbonate vein stockworks from multiple generations of folding. Figure7-7 illustrates the shape of the mineralised quartz-tourmaline altered host unit.

18^th September 2018 Page 60

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure7-7 Cross Section of Gara Fold Structures

Gold mineralisation is strata-bound and hosted predominantly within the quartz-tourmaline stockwork veins, which are enveloped within footwall greywackes and hanging wall sandstone. Higher gold grades values typically occur where the intensity of tourmalisation and stockwork veining are strongest. The sulphide assemblage predominantly consists of disseminated auriferous pyrite with minor chalcopyrite, scheelite, and nickeliferous sulphides.

In the open pit area, the high-grade mineralisation is concentrated along thesub-horizontal fold hinge axes, whereas within the underground area, high-grade mineralisation plunges shallowly southward, parallel to the large scale open warp fold axis . These differential orientations of mineralisation are a result of the earlier, deposit scale warping locally influencing the geometry of the superimposed “S folds” during their formation.

Alteration

Gara is characterised by intense tourmaline alteration. This tourmalisation appears gradational from hanging wall contact to the footwall, its intensity varies from strong to weak. It is also associated with carbonate and silica alteration (veining). Chloritisation is mainly observed in the hanging wall. The sulphides are mainly pyrite (abundant), however arsenopyrite and chalcopyrite have been observed rarely. There is also a correlation between the intensity of tourmalisation, sulphide intensity, silica-carbonate veins, brecciation and gold mineralisation.

Very little saprolite material generated from limited weathering is present at Gara, and as a result the deposit is split into oxide and sulphide.

Mineralisation

Gold mineralisation is strata bound and hosted predominantly within the quartz-tourmaline sandstone unit which is enveloped within footwall greywackes and hanging wall schistose sandstone. Higher gold grades typically occur where the veining is most intense and the range of vein orientation more complex and are mostly in association with carbonate-pyrite. Mineralised veins also contain minor chalcopyrite, scheelite, rare earth element phosphates and nickeliferous sulphides.

18^th September 2018 Page 61

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Baboto

Baboto is a shear hosted deposit situated along a north-south striking shear structure located approximately 14 km NNE from the Yalea deposit. This 5 km long structure contains the Baboto South, Centre and extends to Baboto North on a neighbouring permit.

Baboto is dominated by a thick sequence of metasediments consisting of siltstone, quartzite, greywacke, polymictic conglomerate, and structural breccias with intercalated, strongly sheared argillite. Late cross-cutting mafic intrusives transgress all lithological units.

Structure

The strike of the bedding varies from NNE to WNW and the dip varies from steeply east to steeply west. The main shear zones are vertical to steeply west dipping at Baboto South and sub vertical in Baboto Centre.

Alteration

A very weak tourmaline alteration affecting the medium and coarse-grained greywacke is observed in the extreme NW and east of the target.

Mineralisation

Gold mineralisation is mainly associated with the finely disseminated pyrite occurring in the brittle-ductile shear breccias. These zones generally have a lensoidal shape defined by a series ofsub-parallelN-S shears that follow the lithological contacts between the conglomerate, quartzite, and greywacke. The tabular pyrite or arsenopyrite does not correlate with gold mineralisation, while the fine acicular arsenopyrite is often associated with the mineralised zones. The hanging wall and footwall to the mineralisation are in general strongly foliated rock. The geological model consists of multiple sub parallel dilation structures hosting mineralised zones, which have a lensoidal shape.

Thelow-grade envelope is probably related to the dilation created by the 202° to 220° early sinistral shear. High-grade (4.4 g/t Au to 12.2 g/t Au) en-echelon quartz veins (oriented 240° and dipping 40° to 85° NW and bounded withinN-S trending shears) might be explained by the reactivation of the early 202° to 220° sinistral shear by the later dextral movement.

Loulo 3

Loulo 3 is located 4 km NNE of the Yalea mine. Loulo 3 consists of three mineralised zones: a NNW trending main zone (MZ1) which is situated on the Loulo 3 structure and is transected by the NNE striking main zone (MZ2), which is situated on the Yalea structure, and the third small sub parallel NW striking footwall (FW) zone. The MZ2 mineralisation has an overall strike length of 1.8 km, ranges in thickness between 6 m to 12 m. Both MZ1 and the FW zone have smaller dimensions, being 580 m in strike, 5 m to 7 m thick and 200 m in strike, with 5 m thickness, respectively.

18^th September 2018 Page 62

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Mineralisation consists of a mixture of quartz and hematite veinlets hosted in a zone of silica-carbonate alteration within local tourmaline alteration in the south.

The host stratigraphy comprises an eastwardly younging metasedimentary sequence of quartzite, argillaceous quartzite, and greywacke. The hanging wall contains a 2 m thick, dark green fine-grained quartzite is associated with the mineralisation and is consistently exposed in exploration trenches and is used as a marker unit within the open pit. Footwall rocks are comparatively less competent, with a 10 m thick argillite in contact with a tourmaline greywacke towards the west.

Structure

The host sequence and mineralisation are intruded andcross-cut by two sets ofsub-vertical mafic dykes, striking north and NNE. The NNE trending mafic intrusive has exploited asub-vertical late fault, which has displaced the mineralisation by uplifting the eastern block by between 5 m and 10 m, suggesting a component of shortening. Subparallel NW trending pods of mineralisation are found on a left-hand jog which has dilated due to sinistral movement.

Alteration

Local weak to moderate tourmaline alteration generally decreases in intensity southward along strike. Strong iron oxide alteration and hematite veinlets are seen within the mineralised zone, which increase in intensity proximal to box work veins and fractures.

Mineralisation

Paragenetic studies show Loulo 3 mineralisation contains numerous sulphide phases, with chalcopyrite occurring late in the mineralisation history. which is situated on the Yalea structure, and the third small sub parallel NW striking footwall zone. Mineralisation consists of a mixture of quartz and hematite veinlets hosted in a zone of silica-carbonate alteration within local tourmaline alteration in the south.

The distribution of high-grade zones is controlled by the narrowing of the host stratigraphy package, which focusses strain and fluid flow, causing the hematite rich Yalea Structure to interact with the silica-carbonate Loulo 3 Structure particularly within MZ2. Gold bearing sulphides predominantly consist of pyrite and arsenopyrite, with chalcopyrite occurring as a latenon-gold bearing phase.

Gara West

Gara West is located in the hanging wall sequence approximately 400 m west of the Gara pit and is characterised by predominantly shear and breccia hosted mineralisation within a medium to coarse grained sandstone unit that is variably altered with tourmaline, chlorite, and silica-carbonate. The sandstone hosts four mineralised lodes striking NNE and dipping moderately westward. All mineralised zones aresub-parallel, striking NNE and dipping steeply westward.

18^th September 2018 Page 63

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The host sequence at Gara West forms part of the same folded stratigraphy in which the Gara deposit is developed, and largely consists of sandstones, limestones, argillaceous quartzites, pink quartzite, and a distinctive heterolithic breccia.

Structure

Bedding is generally oriented NS, dipping 50° west however sinistral folding can produce changes in strike and dips that locally aresub-vertical.

Alteration

The medium to coarse grained sandstone has been preferentially altered with tourmaline (and silica-albite) due to the increased porosity of the protolith, relative to the bounding limestone in both the hanging wall and footwall.

Mineralisation

The gold mineralisation is strata bound and hosted in strongly developed quartz carbonate vein arrays as well as associated disseminations and hydrothermal breccias within pink quartzite. The mineralisation exhibits a pinch and swell geometry at the scale of the deposit and have greater continuity along strike than at depth. Higher grades are associated with increasing intensity of quartz-carbonate vein development, carbonate-hematite-goethite alteration, and intensity of brecciation.

7.5 Gounkoto Permit Geology and Mineralisation

Gounkoto

Gounkoto is a shearzone-hosted deposit consisting of an east-dipping mineralised package of fine-grained sedimentary sandstone units termed QR (after the French term “quartz rosé” meaning pink quartzite), with rare limestones. It is bound in the hanging wall by black sandstone and the footwall by west dipping sediment. The deposit is characterised by a stepped geometry, with wider zones of mineralisation generally seen on the NW trending structures and narrower zones on the north-south trending structures. This is believed to be related to dilation across these structures in a strong sinistral strain environment.

The hanging wall of the deposit is characterised by a thick sequence of fine grained ‘black sandstone’ with or without narrow units of limestone and QR (Figure7-8). The mineralisation is generally hosted in a siliceous QR unit and the footwall consists of argillaceous quartzites (SQR) and/or greywacke with narrow mafic intrusives including micro diorites.

18^th September 2018 Page 64

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure7-8 Typical Plan View and West-East Cross Section across Gounkoto MZ1, MZ2 and MZ3

The Gounkoto system is characterised by strong foliation and shear fabrics with different orientations. Foliation in the hanging wall black sandstone is parallel to bedding and strikes NNE and dips 50° to 60° SE. This fabric intensifies to a shear in the immediate hanging wall of the deposit. Fabrics in theQR- or limestone-hosted mineralisation are generally vertical to steep west-dipping. Black sandstone, SQR and greywacke in the footwall exhibit moderately to steeply west-dipping foliation.

There are six main lithologies present at Gounkoto:

● SQR – ‘Schistose Rose Quartzite’. A host of lower grade gold mineralisation comprising finely interbedded calcareous silts and shales with siliceous material.

● Limestone – fine grained sedimentary dolomitic limestones (or more accurately meta-dolomites due to regional metamorphism).

● QR – ‘Rose Quartzite’. Pink quartzite is the main host to mineralisation at Gounkoto. Fine grained variously silty/shaley arkose grits with variable carbonate contents have been altered (albitic and limonitic) during several events to give the pink colour.

● HWQR – Hanging wall QR. Differs from QR in its lack of first stage chlorite-magnetite alteration and secondary hydration/oxidation events.

● Greywacke – fine to medium grained, strongly pelitic feldspathic arkoses. Generally darker due to strong chlorite and tourmaline alteration.

● Intrusives – Intermediate composition cross-cutting dykes.

Structure

Gounkoto is a large NNW trending shear zone, with a complex assemblage of ductile shear breccias, shears, and faults characterised by a stepped geometry, with wider zones of mineralisation generally seen on the NW trending structures and narrower zones on the north-south trending structures. This is believed to be related to dilation across these structures in a strong sinistral strain environment. The mineralisation is generally hosted in a siliceous ‘Rose

18^th September 2018 Page 65

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Quartzite’ (QR) unit. The mineralisation is subdivided based on the structural and lithological characteristics. From north to south, these are:

● Northern Zone: a narrow package of limestone-hosted mineralisation between hanging wall and footwall shear zones striking NNE and dipping steeply to the east. Mineralisation is of intermediate thickness and grade relative to the rest of the deposit.

● Jog Zone: a broad assemblage of repeated stacked gritty QR units, structurally offset from one another creating a stepped geometry, generally striking NNW. Upright flower structures have created an apparent thickening of units, forming wide mineralised zones of high-grade.

● Pinch Zone: a narrow package ofQR-hosted mineralisation between hanging wall and footwall shear zones striking NS and dipping steeply east close to the surface but shallows in dip at depth. The mineralisation is generally narrow andlow-grade.

● Wrench Zone: a broad package ofQR-hosted strongly mineralised lodes between hanging wall and footwall shear zones striking NW and dipping 40° to 50° east.

● Southern Zone: a narrow package ofQR-hosted mineralisation between hanging wall and footwall shear zones striking NS and dipping steeply east. The mineralisation is generally narrower and lower grade than the Northern and Wrench Zones.

At a deposit scale the interaction between older NNE structures and a NNW structure in wrench zone and between the older NNE and NS in north is evident. It appears that the interaction between those structures is responsible for the low angle NE plunging zones of high-grade mineralisation recognised at Gounkoto. These NW structures are either steep strike-slip faults which may have tapped deep fluids, or they are related to a more regional thrusting event. The NS structures are thought to be related to the movement on the Senegal-Malian shear structure during early accretion. The NNE structures are related to the bedding, which has been exploited by the shearing.

Alteration

Several phases of alteration are evident at Gounkoto. Initial greenschist facies regional metamorphism resulted in sericitisation and chloritisation. This was followed by a strong oxidation and hydration event which particularly affected pelitic and fine-grained units, marked by the introduction of limonite into the system. Subsequent phases of hydrothermal activity introduced a pink-orange silica-albite alteration which affected more permeable siliceous QR and SQR units.

Several phases of oxidation occurring at various stages relative to gold mineralisation resulted in iron mineralisation, producing characteristic zones of purple and red oxidised hematite alteration.

Mineralisation

Gold is strongly associated with sulphide mineralisation, dominantly pyrite. Magnetite, chalcopyrite, arsenopyrite and pyrrhotite are also present locally and have a strong association with gold. Gold is also commonly found in gangue in zones of strong silica-carbonate alteration, suggesting that remobilisation also played a role in gold (re)distribution at Gounkoto.

The initial stage of mineralisation involved the deposition of euhedral and equant grains of magnetite in stringer networks, veinlets, and aggregates, and accompanied by chlorite mineralisation. This magnetite mineralisation was subsequently subjected to a strong oxidisation

18^th September 2018 Page 66

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

and hydration event resulting in almost total replacement of magnetite to haematite and annealing of the stringer network together to form aggregate almost massive bodies of haematite, leaving magnetite as small relict grains within haematite.

The onset of the second stage of mineralisation at Gounkoto was synchronous with the oxidation of magnetite and growth of haematite, occurring initially as pyrite (with minor gold and chalcopyrite) which are locally included within the aggregate bodies of haematite. The pyrite mineralisation event was pervasive in nature and was accompanied by strong silica-carbonate alteration, creating the dominant silica-carbonate-pyrite mineral assemblage at Gounkoto. This early pyrite is typically surrounded and enveloped by later stage pyrite with a more cellular and filamentous habit with which gold is commonly associated.

Faraba

The Faraba deposit strikes NNW and is comprised of several zones of gold mineralisation hosted within and along the contacts of north-south striking, coarse grained, gritty sandstone units (lithic wackes) in a package of sheared argillaceous sediments. The host sandstones for mineralisation are arkosic grits comprised of plagioclase and quartz grains in a matrix containing dolomite and clay minerals. Lithologic layering (transposed bedding) dips steeply westward; however, the mineralised zones dip steeply to the east.

The mineralisation terminates where the Faraba Structure meets the argillite units on either side of the sandstones. The resulting mineralisation occurs as numerous silica-carbonate and secondary iron oxide altered sub vertical panels with narrow east-west dimension, each containingsub-horizontal to shallow plunging zones of higher grade.

Structure

The structural model involves theNNW-striking Faraba Structure intersecting favourable (competent) coarse grained sandstone units. The resulting mineralisation occurs as numerous upright, north-south striking panels with narrow east-west dimension, intermediate vertical dimension with a maximum dimension along strike, each containingsub-horizontal to shallow southward plunging zones of higher grade. A consequence of this geometry is that mineralisation cannot be projected down dip on either shear structures or host lithologic layers, which reduces the continuity of mineralisation in cross section. The plunge control on higher grade mineralisation is the result of the intersection between individual shears of the Faraba System and sandstone units within the local stratigraphy. When these structural elements aresub-parallel in strike,sub-horizontal shoots are produced.

As the discordance in strike orientation between the shears and sandstone increases, the higher grade shoots adopt a general southward plunge (up to 25°), although northward plunges are also observed in the system locally. These reversals in plunge are produced by the anastomosing geometry of the shear structures intersecting the planar stratigraphy, which creates intersection lineations that are curvilinear (and locally north plunging).

At depth, aNNW-striking, 60° W dipping fault, with a surface trace located to the east of the Faraba System, truncates the resource at depth with an unknown offset.

18^th September 2018 Page 67

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Alteration

Hydrothermal alteration is multi-phase, with early silica alteration and barren arsenopyrite. Later phases of fluid flow introduced silica-carbonate with subordinate iron alteration (magnetite) associated with pyrite and gold.

Mineralisation

During mineralisation, interpretedNW-directed shearing has produced an array of mineralised east-dipping veins and breccias in the more competent sandstone host rocks. Gold mineralisation is dominantly hosted by pyrite, with local magnetite, chalcopyrite, arsenopyrite and pyrrhotite. Gold is also commonly found in gangue in zones of strong silica-carbonate alteration, suggesting that remobilisation also played a role in gold (re)distribution at Gounkoto.

7.6 Discussion

An independent external audit by QG Consulting (QG) that was completed in early 2015 identified that the broad geological setting at Yalea and Gara is well understood and that they are predictable for the mine geologists.

QG also commented that the mineralisation is easy to define in the underground faces, drill core as the alteration and veining makes the mineralised contact easy to define. QG concluded that at Loulo, the conversion of geological understanding into wireframes supports the Mineral Resource estimation process.

QG found that at Gounkoto, the geology is more complex and that there is still a relatively high degree of uncertainty in the interpretations when compared to Yalea and Gara. QG commented that as the pit gets deeper and the mining transitions to underground, the interpretation carries a greater risk for increasedore-loss and dilution. Randgold has taken steps to counteract this risk at Gounkoto through additional drilling, interpretation modifications and, most significantly, the expansion of the pit to include more material that was previously part of the underground resource.

The QP is confident that Loulo-Gounkoto teams have a clear understanding of the geology and mineralisation so that geologically valid interpretations and wireframes can be created to prepare the Mineral Resource models.

18^th September 2018 Page 68

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

8 Deposit Types

Paleoproterozoic meta-volcanic and metasedimentary rocks, and associated intrusive rocks are collectively referred to as Birimian and are the primary source of gold deposits across West Africa. The Birimian mineralisation occurs primarily along the edges of volcanic belts along faults that cut through and near the host greenstone rocks. The Loulo and Gounkoto deposits can be classified as a typical shear hosted Birimian style mesothermal gold deposits.

The Birimian turbiditic sedimentary rocks have been intruded with multiple sills and dykes of varying composition. The package of Birimian rocks has been influenced the Eburnean orogeny which has subjected the rocks to deformation and metamorphism. These deformation events have resulted in the shearing which hosts multiple gold deposits across the region. Gold mineralisation is associated with shears and intrusions. It is found:

● Either bound between shears,

● Occurring adjacent to a shear,

● Within a favourable lithology that has been structurally prepared by shearing.

The mineralisation is hosted within the volcaniclastic and sedimentary package. These deposits tend to have significant strike and depth potential, with exploration concentrating on delineating strike and depth extent, followed by infill drilling within the zones of better continuity and grade.

Tourmalised greywacke outcrops are used to potentially locate Gara style mineralisation. Structural analysis is used to locate zones of possible dilation that can result in Yalea style mineralisation.

The airborne electromagnetic survey integrated with structural and surface geochemical data has been used to identify resistive and conductive zones which are indicative of mineralisation.

18^th September 2018 Page 69

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

9 Exploration

Immediately after acquisition of the Loulo Project Randgold completed:

● 13,155 m core and 8,996 m RC drilling which led to the discovery of Yalea.

● Validation and infill drilling at Gara with 15 drill holes.

● Sterilisation drilling at proposed mine infrastructure sites.

● Regional exploration programs with remote sensing, geological and regolith mapping, details soil grids along with pitting and trenching.

● Re-interpretation of the BHP soil data.

Since then, Randgold has undertaken numerous RAB, RC and diamond drilling programs along with trenching to target: Gara, Yalea, Loulo 3, Loulo 2, Baboto, Gara West, P129, Faraba, and Gounkoto. Exploration continues at Loulo-Gounkoto to replenish resources that have been depleted from mining.

As the knowledge base grows through continual geological documentation during mining, brownfields exploration is undertaken adjacent to the known mineralisation, in an attempt to locate further extensions. Successful targeting through the reinterpretation and integration of the exploration strategy is demonstrated with the discovery of Gounkoto.

9.1 Exploration Concept

Exploration at Loulo-Gounkoto is focussed on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit targets for high-grade mineralisation based on the structural model. This program of drilling aims to discover additional potentially economic resources at step out distances greater than 400 m from the current limits of the resource block models. Recent geophysical surveys have been integrated with updated geological maps to develop a tectonostratigraphy of the Kofi Series rocks at Loulo-Gounkoto, to improve the understanding of the controls to gold mineralisation and the regional geologic architecture. This Project-wide geologic framework is driving are-assessment of exploration work to date as part of greenfields target generation.

The Loulo district is extremely prospective for gold mineralisation and the tactic is to locate suitable dilatational traps on, or adjacent to the main gold bearing structures that can host significant economic mineralisation. Going forward, fieldwork will include mapping, lithological sampling and trenching to develop and test geologic models for these targets to advance them through the resource triangle on a results driven basis.

The current exploration concept has been proven to be effective, with both the discovery of Gounkoto and the successful replenishment of depleted Mineral Resources and Ore Reserves at both mines.

18^th September 2018 Page 70

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

9.2 Loulo

2017 Loulo Exploration

A Project wide integrated target map at Loulo has been completed incorporating field mapping, satellite imagery, DTM data, soil geochemistry, and regolith interpretation in an integratedre-ranking of targets.

At Yalea, surface resource definition drilling within the Yalea Transfer Zone is designed to test previously observed high-grade intersections returned strong intercepts.

At Gara, drilling continued to target Gara Far South Extended for the purposes of defining an Indicated Mineral Resource and better defining the position of the cataclastic fault with associated hydrothermal albitite that offsets the mineralisation.

At Loulo 3, diamond drilling was used to test the MZ1 and MZ2 strands of the Loulo 3 structure to expand the defined mineralisation.

A structural review of Gara West was completed in 2017 which has significantly upgraded the existing resource by identifying higher grade shear and breccia mineralisation hosted in a tourmaline and albite altered sandstone unit. This mineralisation ranged from 0.8 m to 13.4 m thick and averaged 4.0 m thick across the 68 intersecting holes.

Greenfield field mapping and trenching was completed to provide additional information for the integrated target map.

Planned 2018 Loulo Exploration

The key aims of the 2018 Loulo exploration programme are to replace 2018 depletion and discover a stand-alone resource or significant satellite to increase Mineral Resources and LOM of Loulo-Gounkoto.

Surface resource definition of Yalea Transfer Zone and Yalea Central Deeps that commenced in 2017 is planned to continue into 2018. Further exploration drilling is planned to target the Yalea intersections panel immediately below the Yalea South Transfer Zone target.

Follow up drilling will be used to target potential extensions to the additional Gara West mineralisation intersected in 2017, as the system appears open at depth and along strike to the south.

Loulo 3 drilling is designed to target plunge extensions of MZ2 shoot at near surface and at depth and to test the variability in the mineralisation thickness and grades across at a range of drill spacings.

A structural review will be completed on the entire Gara West system to determine the potential of higher grade mineralisation remains open at depth and along strike.

18^th September 2018 Page 71

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

9.3 Gounkoto

2017 Gounkoto Exploration

As at Loulo, a Project-wide integrated target map encompassing field mapping, satellite As at Loulo, a Project -wide integrated target map encompassing field mapping, satellite imagery, DTM data, soil geochemistry, and regolith interpretation map was drafted tore-rank targets for exploration.

Exploration focussed on structural modelling and drill design to test for potential extensions to the mineralisation in MZ4, MZ3 Main, and P64E, as well as in the footwall of MZ1 in order to reduce the strip ratio of the current ‘Super Pit’ shell design.

Re-logging and sectional interpretations of the Iron Structure within Gounkoto deposit has validated high-grade mineralisation hosted on three individual shears to establish the potential viability of extending the Gounkoto pit to include this mineralisation.

Several greenfield targets located along major prospective structures (domain boundary and Faraba Structure) are also being advanced for additional satellite resource potential.

Planned 2018 Gounkoto Exploration

In the same manner as Loulo, the Gounkoto exploration is focused on the primary objective of replacing 2018 depletion and discovering additional resources to increase the inventory of Loulo-Gounkoto.

During 2018, exploration will complete afollow-up RC drilling programme on Faraba North target to define potential extension of the Faraba system. Additionally, the potential extension of a hematite structure, in the eastern portion of the Faraba Main open pit will also be tested.

A gap analysis across the entire Gounkoto System has identified opportunities along a 250 m strike length of the Domain Boundary. A twin diamond drill hole has been approved to better understand how the interpreted footwall-finger style mineralisation relates to the HW Domain Boundary in Faraba. Additional drilling will target the domain boundary plunge.

9.4 Discussion

Loulo and Gounkoto have a detailed Standard Operating Procedure (SOP) Manual for Exploration and Drilling Practices that provides standardisation and consistency for all field technical personnel to ensure the collection of quality data. The Exploration Manager and Mineral Resource Manager are very experienced in the region and deposit style.

The 2018 planned exploration targets both potential extensions of mineralisation will provide crucial information to further geological knowledge across the complex. The integrated target map update andre-ranking of targets within Loulo and Gounkoto permits that was undertaken in late 2017 is a crucial study that should provide an effective plan for long term regeneration of depleted Mineral Resources.

18^th September 2018 Page 72

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

10 Drilling

RAB, RC, and diamond drilling have been used at Loulo and Gounkoto, although RAB drilling is not used for Mineral Resource estimation. Outcrop, soil, and trench samples are also used for early stage exploration. Selected trench data is used for resource estimation, and the sampling parameters are the same as utilised for RC drilling.

An example of the drilling density for each deposit is presented in section 14.12 Resource Classification.

A February 2015 Independent Mineral Resource audit by QG deemed that the data collection procedures follow industry standard practices.

10.1 Drill Hole Summary

All soil sampling, mapping, trenching, and geological supervision of drilling has been conducted by Randgold geologists. Table10-1 and Table10-2 presents the known drilling by year, company, and type at Loulo and Gounkoto Permits. An example of the drill density by deposit is detailed in Section 14.14Block Model Validation.

Table10-1 Loulo Drilling Summary

Year	Company	Diamond		RC		RAB		Trench		Channel		Total
		Hole Count	Meters	Hole Count	Meters	Hole Count	Meters	Hole Count	Meters	Hole Count	Meters	Hole Count	Meters
1993	BRGM	125	15,057	—	—	—	—	—	—	—	—	125	15,057
1994	BRGM	30	1,790	—	—	—	—	4	86	—	—	34	1,876
1994	BHP	50	6,634	—	—	—	—	—	—	—	—	50	6,634
1996	BHP	5	620	—	—	—	—	—	—	—	—	5	620
1997	BHP	15	2,335	—	—	—	—	—	—	—	—	15	2,335
1997	Randgold	95	15,358	—	—	915	17,916	—	—	—	—	1,010	33,274
1998	Randgold	37	4,803	—	—	267	4,929	—	—	—	—	304	9,732
2000	Randgold	21	3,640	—	—	—	—	—	—	—	—	21	3,640
2001	Randgold	16	1,270	—	—	255	6,855	16	667	—	—	287	8,792
2002	Randgold	—	—	—	—	—	—	49	988	—	—	49	988
2003	Randgold	84	18,244	—	—	—	—	317	14,101	—	—	401	32,345
2004	Randgold	84	31,962	77	6,131	719	22,409	25	1,342	—	—	905	61,844
2005	Randgold	124	63,565	135	7,752	145	5,332	83	3,753	—	—	487	80,402
2006	Randgold	35	17,606	884	31,456	371	7,442	1	182	—	—	1,291	56,686
2007	Randgold	52	26,677	1,283	44,395	1,652	38,751	27	1,205	—	—	3,014	111,028
2008	Randgold	88	8,092	763	25,548	—	—	104	4,950	127	1,745	1,082	40,335
2009	Randgold	97	11,677	3,474	131,213	—	—	—	—	167	1,618	3,738	144,508
2010	Randgold	138	13,255	834	33,203	4	86	62	3,287	165	1,601	1,203	51,432
2011	Randgold	245	35,748	346	19,882	—	—	—	—	431	3,279	1,022	58,909
2012	Randgold	206	29,772	733	31,301	—	—	1	22	519	3,278	1,459	64,373
2013	Randgold	145	26,718	42	3,838	—	—	33	2,028	757	5,093	977	37,677
2014	Randgold	319	65,644	2	228	—	—	77	6,566	526	4,090	924	76,528
2015	Randgold	209	58,185	81	5,061	—	—	63	4,635	790	5,352	1,143	73,233
2016	Randgold	365	90,219	1,644	64,511	—	—	28	2,113	752	4,547	2,789	161,390
2017	Randgold	450	98,792	163	9,493	—	—	41	3,316	833	5,222	1,487	116,823
Total		3,035	647,663	10,461	414,012	4,328	103,720	931	49,241	5,067	35,825	23,822	1,250,460

18^th September 2018 Page 73

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table10-2 Gounkoto Drilling Summary

Year	Company	Diamond		RC		RAB		Trench		Total
		Hole Count	Meters	Hole Count	Meters	Hole Count	Meters	Hole Count	Meters	Hole Count	Meters
1993	BRGM	15	1,290	15	852	—	—	—	—	30	2,142
2000	Randgold	—	—	2	200	—	—	—	—	2	200
2005	Randgold	—	—	31	3,943	165	6,813	26	1,836	222	12,592
2006	Randgold	8	2,000	36	3,092	2	60	54	3,499	100	8,651
2007	Randgold	15	3,694	—	—	346	10,152	2	65	363	13,911
2008	Randgold	11	2,992	—	—	398	10,432	—	—	409	13,424
2009	Randgold	79	16,733	32	2,921	175	4,953	12	594	298	25,201
2010	Randgold	140	46,941	627	53,380	—	—	17	816	784	101,137
2011	Randgold	70	34,925	222	18,097	—	—	105	8,721	397	61,743
2012	Randgold	12	6,990	1,345	39,775	—	—	7	458	1,364	47,223
2013	Randgold	62	22,999	930	57,366	—	—	22	1,682	1,014	82,047
2014	Randgold	34	14,884	448	35,064	—	—	47	3,097	529	53,045
2015	Randgold	17	6,108	1,061	58,300	—	—	4	238	1,082	64,646
2016	Randgold	45	14,993	1,050	59,079	—	—	15	1,012	1,110	75,084
2017	Randgold	17	4,364	965	52,210	—	—	48	4,148	1,030	60,722
Total		525	178,913	6,764	384,279	1,086	32,410	359	26,166	8,734	621,768

10.2 Survey Grids

Loulo-Gounkoto uses UTM WSG84 Zone 29 grid for all surveys.

10.3 Drill Planning and Site Preparation

Drill holes are planned in Vulcan and Micromine software. Consideration is given to the orientation of the drilling in relation to the geological structures, to provide for unbiased sampling. Surface drill holes are principally planned to intersect the mineralisation perpendicular to the main body of mineralisation. The planned collar location is marked by the Mine Surveyor using a differential GPS (DGPS). If required, the drill pad is cleared around the collar marker to ensure sufficient room is available for the drill rig, auxiliary vehicles, and sample collection.

Underground drill collars, as well as back sights and foresights, are surveyed using total station underground survey instruments, and marked on the drift walls, by the Loulo Mine surveyors.

If the drill hole is to be completed using diamond drilling, then a sump is dug in one corner of the pad to collect the drilling returns. The sump is fenced off with a temporary barrier and security tape.

The Senior Geologists, Drill Contractor, Mine Planner, Mine Surveyor and Mineral Resource Manager all sign off on the drill hole plan prior to initiating drilling.

There are three categories of drilling at Loulo-Gounkoto:

• Exploration Drilling – wide spaced exploratory and resource definition drilling. This category will include diamond drilling with RCpre-collars (RC_DDH).

• Advanced Grade Control Drilling – consists of wider spaced drilling to position underground Footwall (FW) drives and for Mineral Resource upgrades in the open pits.

• Infill Grade Control Drilling – used for final production definition for Measured Mineral Resources / Proved Ore Reserves. Generally, Loulo-Gounkoto inventory of Infill Grade Control Drilling is some three to six months inventory for open pit and approximately 18 months for underground.

18^th September 2018 Page 74

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

10.4 Downhole Surveying

All drill holes are surveyed down-hole using a multi shot Reflex EZ-Trac or a conventional Gyro surveying tool which measures the azimuth and dip every 25 m down hole. Drill holes are also surveyed as drilling is ongoing using a single shot measurement to ensure the hole is progressing as planned. The Gyro is used in any zones with magnetic interference. Some drill holes that were drilled for the underground feasibility study were measured with the Gyro on 5 m intervals. During 2018, the EZ-Trac will be replaced with anon-magnetic north seeking Reflex Gyro.

The survey measurements are manually entered into the database. With the planned upgrade to theEZ-Gyro tool the surveys will be stored on a handheld Panasonic controller which will connect to the Reflex Hub Management software and upload the surveys directly to the Datashed database, thereby removing any potential human errors during transcription.

The Reflex down hole survey tools are returned to the manufacturer for calibration on an annual basis. A known survey tool test stand has been erected at the Project so that the tools can be checked weekly.

10.5 Collar Surveys

Once a drill hole, or surface trench, is completed and all machinery has moved away from the site, the drill hole is surveyed using a DGPS for surface drilling and by total station underground by Randgold Surveyors.

The drill collar is plugged, and a concrete collar surround is placed which is labelled with the HoleID and survey date. The sumps are backfilled, and the drill site is remediated. Trenches are surveyed by DGPS starting at the collar (initial cut) and then additional point measurements are taken every 2 m.

10.6 Diamond Drilling

Randgold primarily uses Boart Longyear Canada for underground and surface drilling. Initial drilling at Loulo was completed by West African Drilling Services. Some minor Gounkoto grade control drilling is completed by a local contractor, DCS Mali SARL. Core sizes are HQ (63.3 mm core diameter) in saprolite (oxides) and NQ (47.6 mm core diameter) in the unweathered rock. On completion of drilling, all diamond holes are photographed.

Core recoveries are generally good, with an average of 97.4% recovery in the unweathered rock, 84.6% recovery in the transitional zone and 76.0% in the saprolite zone. Average recovery in the mineralised zones was 97.6% with a range of between 80% and 100%.

Drilling Procedure

A project geologist must be on site prior to drilling commencing and ensure that the drill rig is lined up as per the drill plan as well as supervising drilling, core orientation and down hole surveying. Once each drilling run is complete, the drill core is removed from the drill rod and

18^th September 2018 Page 75

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

placed in an angle iron rack to mark up an orientation line with red chinagraph pencil or crayon from the data received from a Reflex ACT II Core Orientation Tool (ACT). The apex of the structure is marked on the core in a chinagraph pencil or crayon by the core technician. If the orientation and apex lines are overlapping, then the apex line is offset by 5 mm.

Diamond core is transferred to the core trays and a plastic down hole depth marker is placed at the beginning and end of each core run with the depth marked on it. All areas of core loss are identified, and the core is marked up for core recovery. Each drill core box is marked with the Hole ID, top and bottom depth of the core and the box number. The core is then transferred to the core yard facility by Randgold staff for logging and sampling.

Core Logging

Diamond drill core is logged geologically including, weathering, mineralisation, alteration, structure, lithology, and redox. Core recovery is measured both in the field and during detailed logging. All structural data including alpha and beta angles of any structures are also recorded at this stage. Logging is completed on to paper logs by the geologist which are transcribed to the central database after verification.

All diamond drill core is oriented and where determining orientation is not possible the core is assembled with previous runs, where possible, to try and extend the orientation line, such that structural directions in the form of alpha and beta angles are documented.

Geotechnical logging is only performed on holes specifically drilled for geotechnical assessment. Diamond core is photographed both wet and dry to provide a safeguard for core loss and to aid geological interpretation.

Sampling

Diamond drill half core samples are taken between 0.8 and 1.2 m long and split at boundaries between geological units. The drill core is split using a diamond saw with fresh water. Half core is submitted for assay wherever possible, quarter core was historically only used when there is a need for further assays or other analysis to be completed which need to be completed on the same interval. Metallurgical sampling is generally only undertaken on either dedicated metallurgical drill holes or on coarse crush duplicates.

Diamond core is photographed both wet and dry before being halved with a diamond saw. One half is submitted for assay analysis whilst the other half is stored for future reference.

10.7 Reverse Circulation Drilling

RC drilling has been used historically in the open pit but is no longer used at Yalea and Gara underground operations. RC drilling is still used at other open pit operations and brownfield exploration. RC drilling is used at Gounkoto for grade control and exploration/resource development.

18^th September 2018 Page 76

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

RC drill holes to date have been drilled at a 5.5 inch / 140 mm diameter which produces approximately 25 kg of material per 1 m sample interval. The RC drill rods are 3 m in length.

Drilling Procedures

Most RC samples are passed directly through a Gilson splitter. Alternatively, a cone splitter is used on some rigs which are utilised less frequently. RC samples are collected onone-meter intervals. RC chips are sieved and logged by the geologist before being placed in chip boxes for storage.

A Project geologist is present during all RC drilling. Prior to commencement of any drilling he/she will print out cross sections for all planned holes, indicating the expected geology and mineralisation to be intersected and ensure that prepared and labelled sample bags are present. The geologist is responsible for the site layout to ensure that, where possible, the drilling and sampling operations will not interfere with each other and that the sampling is not taking place down wind of the drilling.

A 6 m PVC casing is used to collar the hole to help prevent drill hole collapse and sample contamination. Once the drill hole has been collared the geologist ensures that the drill hole is cleaned to remove any material that may have ingressed during collaring before drilling is resumed.

If the drill hole intersects the water table an auxiliary booster(s) is used to ensure that the samples are dry. After each rod change air is blown down the hole prior to recommencing drilling to dry it out.

If the sample is moist or damp, the bulk sample is weighed and recorded as a “wet sample weight”. The moist or damp samples are transported to the sampling area and dried. Once dry the sample isre-weighed and the dry weight recorded in addition to the previously recorded wet weight. This is rarely an issue due to mine Permit dewatering infrastructure.

Reverse Circulation Logging

All RC samples are logged on 1 m intervals as per the sample length received from the RC rig. time. For each drill hole lithology, visible mineralisation, vein intensity, alteration, oxidisation, and depth of water table are logged as a minimum.

RC sample recovery is measured by weighting the total weight of sample collected over eachone-meter interval and comparing this to the theoretical expected weight for the lithological unit and weathering type.

Sampling and Splitting Procedure

A 12.5 kg split sample is produced from the initial 25 kg RC sample through the primary Gilson splitter. This 12.5 kg samples are then passed through a second Gilson splitter to produce a 3 kg to 4 kg sample which can be submitted for analysis.

18^th September 2018 Page 77

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

After splitting the sample, both the primary sample and rejects are weighed and recorded. A small amount of material is taken from the sample rejects and placed in a slotted plastic rock chip sample tray, which arepre-labelled with the drill hole ID and sequence numbered drill hole intervals in each sample slot. RC sample recovery is measured by weighing the total weight of sample collected overone-meter sample and comparing this to the theoretical expected weight for the lithological unit and weathering type.

The RC samples are believed to be representative of mineralisation present at all Loulo and Gounkoto Mineral Resources. The QP considers the quality of RC drilling to be at an industry accepted standard.

10.8 Trenching

Most of the trenches were excavated perpendicular to strike of the known mineralisation. The trenches are excavated through all transported material until at least 50 cm of insitu saprolite is exposed at the base of the trench. All trenches were geologically mapped in detail by a geologist on both walls. As part of the mapping the lithology, alteration, mineralogy, and orientation of any visible structures are recorded.

The standard sampling procedure was such that a shallow cut, approximately 15 cm to 20 cm wide channel is excavated at the bottom of the trench and sampled on geological and mineralisation contacts. The dimensions of the channel are kept constant through the entire sample length.

The sample is split onsite using a Gilson splitter and a 3 kg representative sample placed in cloth sample bags with a sample ticket and sent to the laboratory.

10.9 Other Sampling Methods

RAB and pitting has been used at Loulo-Gounkoto for exploration purposes however it is not used for resource estimation purposes.

10.10 Drill Twinning Studies

Twin drilling studies have been undertaken at Yalea and Gara as part of the original feasibility studies prior to construction of underground mine development. These comparisons have shown that although there can be variations in grade, as expected, the broad intercepts and relative grade of the intersections are comparable across the twins.

Twin drilling has also been undertaken at the Gounkoto MZ and hanging wall. The results of have shown that, although there are local significant grade variations, the broad intercepts and relative grades are generally relatively comparable across the twin intercepts. The FW finger zone, however, has shown dramatic variations on very closely spaced (less than 5 m) drill holes.

The variation in the drill twinning studies has been used to feed the classification criteria for the Mineral Resources. The FW finger zone at Gounkoto, which has the dramatic short scale

18^th September 2018 Page 78

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

variations in grade, has a much tighter drill spacing for Measured Resources (6 m by 6 m) when compared to the remainder of Gounkoto (12.5 m by 12.5 m).

10.11 Discussion

Resource infill and grade control drilling is resulting in local changes to the geological interpretation and modelled grade.

In the QP’s opinion, the drilling and sampling procedures at Loulo-Gounkoto are robust, suitable for the style of mineralisation and are at or above industry standard practices. Regular reviews, external audits and training are undertaken to ensure that this remains so.

The average drill core recovery is 97.4% in the unweathered rock, 84.6% in the transitional zone and 76.0% in saprolite zone. The average recovery in the mineralisation was 97.6% with a range of between 80% and 100%. There are no drilling, sampling or recovery factors that could materially impact the accuracy and reliability of the results.

18^th September 2018 Page 79

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

11 Sample Preparation, Analyses and Security

11.1 Sample Selection

The sample boundaries of drill core are determined based on geology and alteration and most often varies from 0.8 m to 1.2 m. Half core is used whenever possible. Historically, quarter core was only used when there is a requirement for a duplicate assay or if another analysis type was required for the same interval. More recently metallurgical samples are taken from dedicated metallurgical drill holes or from coarse reject material.

RC samples are collected on the rig inone-meter intervals and passed through two Gilson splitters to create 2 kg to 3 kg samples. Wet samples are dried before being split.

11.2 Sample Preparation

All samples submitted for analysis are prepared and analysed at the independent SGS Loulo laboratory, which is managed and self-certified by SGS and located on the Loulo mine site.

Grade control and exploration drill samples are prepared in the same manner. Once the samples are received by SGS Loulo, the sample is weighted and entered into a LIMS tracking system. Samples are dried in an oven at 105°C. Channel and trench samples are disaggregated to remove dry lumps. Dried samples are crushed to ensure that 75% of the sample is below 2 mm.

The crushed sample is then passed through a riffle splitter and the reject material is retained. The 1.5 kg split sample is pulverised in an LM2 pulveriser until 85% passes through a75-micron (200 mesh) screen and a 200 g is split removed and placed in a packet. The LM2 pulveriser is cleaned with an air hose every sample, and with blank material every 6^th sample. SGS Loulo undertakes regular screen sieve tests on the crushing and pulverising. The coarse (6 mm) reject and the pulp (75 micron) reject material are returned to Randgold for storage at the mine site and futurere-analysis if required.

An external audit completed by QG in February 2015 concluded that the sample preparation procedures followed standard industry practices.

The QP concludes that the sample preparation procedures are regularly checked and follow industry standard practices.

18^th September 2018 Page 80

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-1 outlines the flowsheet for diamond drill core sample preparation.

Figure11-1 Summary of Diamond Core Sample Preparation Flowchart - Exploration and Grade Control

18^th September 2018 Page 81

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-2 outlines the flowsheet for RC, channel, and trench sample preparation.

Figure11-2 Summary of RC, Channel, and Trench Sample Preparation Flowchart - Exploration and Grade Control

18^th September 2018 Page 82

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-3 outlines the flowsheet for sample preparation at SGS Loulo.

Figure11-3 SGS Loulo Laboratory – Summary of Sample Preparation Procedure

18^th September 2018 Page 83

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

11.3 Sample Analysis

All Samples for Loulo-Gounkoto are analysed at the SGS Loulo laboratory located on the mine site. SGS Bamako is used for sample overflow and analysis that could not be completed at SGS Loulo. Both laboratories are operated independently and self-certified by SGS.

All samples are analysed by fire assay (FAA550) with atomic absorption finish which has a detection range of 0.01 g/t Au to 100 g/t Au. A 50 g sample is split from the pulp and fire assayed. Figure11-4 outlines the flow chart for analysis of samples at SGS Loulo.

In 2017, a total of 40,690 samples were analysed at SGS Loulo from the Loulo Permit (Table11-1), and 61,639 samples were analysed at SGS Loulo from Gounkoto Permit (Table11-2).

Table11-1 Summary of Certified Reference Materials Used at Yalea and Gara

Loulo
Sample Type	Number of Samples	Percentage
DDH	17,197	42%
RC	10,391	26%
Trench/Channel	7,601	19%
Subtotal	35,189	86%
Standards	2,474	6%
Blanks	2,271	6%
Duplicates	756	2%
Subtotal	5,501	14%
Total	40,690	100%

Table11-2 Summary of Certified Reference Materials Used at Yalea and Gara

Gounkoto
Sample Type	Number of Samples	Percentage
DDH	3,366	5%
RC	47,978	78%
Trench	1,950	3%
Subtotal	53,294	86%
Standards	3,531	6%
Blanks	3,360	5%
Duplicates	1,454	2%
Subtotal	8,345	14%
Total	61,639	100%

18^th September 2018 Page 84

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-4 SGS Loulo Laboratory – Summary of Fire Assay (FAA505) Procedure

18^th September 2018 Page 85

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

11.4 Quality Assurance and Quality Control

To ensure that the assay results are reliable, Loulo-Gounkoto has a robust quality assurance and quality control (QA/QC) system in place to minimise errors at each stage and procedures to be followed when errors are identified.

Quality Assurance (QA) is to demonstrate that the sampling and analytical protocols are appropriate and optimal for the deposit in question. It should entail orientation sampling studies and statistical analysis so that appropriate systems and standards can be tailored to achieve quality results throughout all the stages of collecting and analysing data. Ideally, orientation studies are performed at the beginning of or early stages of project evaluation. Setting up systems and standards to ensure quality throughout all of the stages used to collect and analyse data

Quality Control (QC) is a real-time monitoring and analysis to ensure the protocols developed in QA are being adhered to and are returning precise and accurate results. Entails additional sampling and analysis and statistical examination (such as scatter plots, QQ plots etc.).

The insertion rates for all QA/QC samples is set at 5%, apart from field duplicates which are inserted at 2.8%. The field duplicate to normal sample ratio will be reviewed with an intention to increase the insertion rate to match that of other QA/QC samples in 2018. Generally, blanks and field duplicates are only inserted into the sample stream only within mineralised zones in accordance with Randgold’s standard operating procedure (SOP). SGS Loulo also insert their own QA/QC samples into the samples stream as an internal test.

General QA/QC performance improved in 2017 due to strict adherence to sampling procedure and increased supervision at all levels. Certified reference materials (CRM) sourced from ORE Research & Exploration Pty Ltd (OREAS) improved testing of laboratory accuracy as the samples werepre-packaged in 50 g samples ready for insertion.

The level of performance shows that the SGS Loulo laboratory meets industry standards.

All laboratories undertake their own internal QA/QC which includes blanks, duplicates, and CRMs, which are reported to Randgold along side the field sample results. These results of the laboratory internal QA/QC is reviewed separately by Randgold but are not reported below.

Loulo

This report covers QA/QC results from 1^st January to 21^st November 2017 for Gara; 1^st January to 20^th October 2017 for Yalea, and 1^st January to 31^st May 2017 for Baboto which covers all new data used within the Gara, Yalea and Baboto 2017 resource model updates. The other deposits at Loulo were not sampled during the review period, however, all sampling and analysis at Loulo follow the same parameters. Previous QA/QC reporting periods have not been observed to contain any significant sources of error or bias which would have a material effect on the Mineral Resource.

Screen tests for both Gara, Yalea and Baboto samples indicates that over 90% of all samples analysed during the period attain 75 microns.

18^th September 2018 Page 86

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

QA/QC sample ratio was reviewed during the period. The review saw CRM sample insertion at 1 in 18 (5.6%), blank sample insertion at 1 in 18 (5.6%) and field duplicate sample insertion at 1 in 36 (2.7%). At Loulo, 370 pulp duplicate samples of normal samples from SGS Loulo were submitted to AMTEL laboratory, Canada, for umpire laboratory analysis.

Gounkoto

This section covers the QA/QC program from 31^st August 2016, to 14^th July 2017, which includes all new data used within the year end 2017 resource model. Faraba was not sampled during the review period, however, all sampling and analysis at Gounkoto follow the same parameters. Previous QA/QC reporting periods have not been observed to contain any significant sources of error or bias which would have a material effect on the Mineral Resource.

The QA/QC sample ratio during the reporting period showed CRM sample insertion at 1 in 18 (5.6%), blank sample insertion at 1 in 18 (5.6%) and field duplicate sample insertion at 1 in 36 (2.8%). The field duplicate to normal sample ratio will be reviewed with an intention to increase the insertion rate to match that of other QA/QC samples in 2018. At Gounkoto, 261 pulps duplicate samples of normal samples from SGS Loulo were submitted to AMTEL laboratory, Canada, for umpire laboratory analysis.

Certified Reference Materials

Certified Reference Materials (CRM’s) are inserted into batches at a frequency of 1 in 18 representing 5.6% samples to validate results reported by the laboratory and monitor the control and calibration of the instruments used by the laboratory.

All CRMs used in the review period are sourced from Ore Research and Exploration Pty Ltd, Australia, and are oxide or sulphide type with a matrix of feldspar minerals, basalt, and iron pyrites. CRMs are purchased inpre-packaged 50 g samples that require no preparation before being submitted to the laboratory. Asub-set of the total CRMs available are used and are rotated on a quarterly basis to prevent laboratory identification.

CRM results are monitored and classified as a failure if one sample point falls outside of three standard deviations from the certified mean, or three consecutive samples fall outside of two standard deviations (on the same side) of the mean.

CRM results that have a failure outside of three standard deviations are checked for possible CRM swaps. This is investigated by comparing the returned assay grade to the list of known CRM grades values. The CRM samples are supplied by OREAS with CRM ID printed on the bag. This printed ID is photographed during CRM insertion and then removed prior to submission of the CRM to the laboratory. This CRM photograph is used to help identify CRM swaps. A normal sample swap is also investigated to check if a normal drill sample has been labelled as CRM.

In addition to the CRM photographs, swaps can be investigated the technician’s sampling plan document, verifying used sample numbers, reviewing the sample booklet, and comparing against the other CRMs in the batch.

18^th September 2018 Page 87

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

When all the above investigations are complete, and it has been established that a failure has occurred, the following actions are initiated:

● When two or more CRMs failed in batch and the failure is as result of sample swap, the entire batch is called forre-assay.

● When one or more CRM failed in batch and the failure is not as result of sample swap, the entire batch is called forre-assay.

Based on the above controls when a batch isre-assayed and fails again, the samples are flagged but committed into the database whilst new samples are prepared forre-analysis. If a CRM is observed as repeatedly failed over a period of time, then it is removed from storage and is no longer inserted into the samples stream

Loulo

A total of 731 CRMs consisting of 12 different types of certified reference materials were used within the period. Table11-3 and Table11-3 outline the CRMs analysed for Gara and Yalea during the QA/QC reporting period. Figure11-5 and Figure11-6 present tram lines plots for Gara and Yalea which illustrate that no CRMs returned values outside of three standard deviations of the mean during the reporting period.

The bivariate statistics for Gara and Yalea CRM samples analysed at SGS Loulo during the reporting period show that there is a very high correlation between the assay pairs and the expected grades (Table11-4).

Table11-3 Summary of Certified Reference Materials Used at Yalea and Gara

OREAS CRM ID	Certified Au Value (g/t)	+3SD (g/t)	-3SD (g/t)	Yalea			Gara
				Min Assay (g/t)	Max Assay (g/t)	Quantity	Min Assay (g/t)	Max Assay (g/t)	Quantity
OREAS 202	0.752	0.83	0.674	0.72	0.82	75	0.74	0.78	45
OREAS 203	0.871	0.961	0.781	0.84	0.9	21	0.84	0.9	32
OREAS 205	1.244	1.403	1.085	1.22	1.28	40	1.22	1.29	19
OREAS 206	2.197	2.44	1.954	2.18	2.32	20	2.18	2.3	29
OREAS 208	9.248	10.562	7.934	9.22	9.4	46
OREAS 210	5.49	5.946	5.034	5.43	5.66	25	5.44	5.62	22
OREAS 215	3.54	3.831	3.249	3.42	3.64	96	3.42	3.6	41
OREAS 216	6.66	7.125	6.195	6.5	6.74	53	6.48	6.7	34
OREAS 221	1.06	1.168	0.952	1	1.1	23	1	1.1	15
OREAS 222	1.22	1.319	1.121	1.18	1.28	40	1.2	1.28	30
OREAS 228	8.73	9.567	7.893				8.54	8.72	11
OREAS 229	12.11	12.728	11.492				12	12.3	14
Total						439			292

Table11-4 Bivariate Statistics of Gara and Yalea CRM’s Assayed by SGS Loulo During the Period

Summary Bivariate Statistics	Gara	Yalea
Correlation	99.99%	99.99%
Slope	99.84%	99.70%
R2	99.98%	99.98%
Y Intercept (g/t Au)	0.0064	0.0092

18^th September 2018 Page 88

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-5 Tram Line of Gara CRM’s Assayed by SGS Loulo During the Reporting Period

18^th September 2018 Page 89

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-6 Tram Line of Yalea CRM’s Assayed by SGS Loulo During the Reporting Period

18^th September 2018 Page 90

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

A total of 2,172 CRMs, consisting of 10 different types of standards, submitted to SGS Loulo during the review period(Table 11-5).Table 11-6 andFigure 11-7 indicate that the CRMs performed well during the review period.

Table11-5 Summary of Gounkoto Certified Reference Materials Used During the Period

OREAS CRM ID	Certified Au Value (g/t)	+3SD (g/t)	-3SD (g/t)	Gounkoto
				Min Assay (g/t)	Max Assay (g/t)	Quantity
OREAS 202	0.752	0.83	0.674	0.72	0.78	45
OREAS 203	0.871	0.961	0.781	0.84	0.93	512
OREAS 205	1.244	1.403	1.085	1.07	1.3	430
OREAS 206	2.197	2.44	1.954	2.16	2.38	427
OREAS 208	9.248	10.562	7.934	9.14	9.6	514
OREAS 210	5.49	5.946	5.034	5.38	5.7	56
OREAS 215	3.54	3.831	3.249	3.4	3.61	43
OREAS 216	6.66	7.125	6.195	6.64	6.98	65
OREAS 221	1.06	1.168	0.952	1.0	1.1	38
OREAS 229	12.11	12.728	11.492	12	12.4	42
Total						2,172

Table11-6 Bivariate Statistics for Gounkoto CRMs Assayed by SGS Loulo

Summary Bivariate Statistics	Value
Correlation	99.99%
Slope	100.06%
R2	99.99%
Y Intercept (g/t Au)	0.0104

18^th September 2018 Page 91

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-7 Tram line Plot of all Gounkoto CRMs Assayed by SGS Loulo

18^th September 2018 Page 92

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Blanks

Blank samples are free media(Au-free for these analyses) assayed to help ensure no false-positives are obtained from the laboratories and to check for contamination. These samples return gold assay below the analytical detection limit (i.e. <0.01 g/t Au). Previously blanks used for the Project were prepared onsite from barren sandstone material that was sourced approximately 20 km from Gounkoto. Since 2016, commercially sourced OREAS certified coarse blanks have been used.

During the collection of samples, blank sample materials were inserted into sample stream at a rate of approximately 1:18 representing 5.6% of the total sample. These samples undergo the same sample preparation as the drill samples and used to detect inter-contamination due to poor cleaning of sample preparation equipment throughout the varioussub-sampling processes.

Loulo

A total of 730 blank samples were used during the period. All samples were submitted to SGS Loulo. 439 blank samples were analysed for Gara and 291 were analysed for Yalea.

All blanks were evaluated against five time the detection limit as the failure limit. The overall performance shows a 100% pass rate of blank samples assayed (Table11-7).

In the QP’s opinion, the blank samples are performance very good.

Table11-7 Loulo Blank Results Returned During Reporting Period

Type	Min Assay (g/t)	Max Assay (g/t)	N° Samples	Above 5x Detection Limit	Between DL and TL	Below DL	% Pass	% Fail
Yalea	0.005	0.02	291	0	1 (0.34%)	290 (99.66%)	291 (100%)	0
Gara	0.005	0.02	439	0	7 (1.59%)	432 (98.41%)	439 (100%)	0

Gounkoto

A total of 2,170 Gounkoto blanks were submitted to SGS Loulo. All blanks were evaluated against five times the detection as the failure limit. The overall performance shows that 99.95% of the blanks assayed within failure limits (Table11-8). Generally, this level of performance is deemed as good.

Table11-8 Gounkoto Blank Results Returned During Reporting Period

Type	Min Assay (g/t)	Max Assay (g/t)	N° Samples	Above 5x Detection Limit	Between DL and TL	Below DL	% Pass	% Fail
Blank	0.005	0.06	2,170	1 (0.05%)	14 (0.65%)	2155 (99.30%)	2169 (99.95%)	1 (0.05%)

18^th September 2018 Page 93

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Duplicates

Duplicate samples are used to check the homogeneity of the samples prior to sample splitting along with the accuracy and repeatability of sampling, splitting, and assaying At Loulo-Gounkoto, field duplicates are currently inserted every 36 samples. The field duplicate to normal sample ratio will be reviewed with an intention to increase the insertion rate to match that of other QA/QC samples in 2018.

Duplicate samples can be obtained from three sources

● Field Duplicates are obtained from the initial splitting of the RC sample during sampling at the rig

● Coarse (Reject) Duplicates are obtained from the coarse reject sample that is returned from the laboratory after the initial crush to 6 mm. of the entire half core sample.

● Pulp Duplicates are obtained from the pulverised75-micron sample that is returned from the laboratory after the pulp is removed for analysis.

Field duplicates are not undertaken on diamond drill samples as the variance between the two half of core, due to the nuggety nature of gold mineralisation.

Loulo

A total of 281 samples werere-sampled from course crush reject material and were analysed by SGS Loulo during the period. 155 samples were analysed from Gara, while 126 samples were analysed from Yalea.

Duplicate results from the laboratory show a good correlation between original samples at both Gara and Yalea, such that it can be concluded that there is little or no bias in the results (Table11-9).

Table11-9 Bivariate Statistics for Log Scatter of Gara and Yalea Course Crush Rejects Duplicate Assayed by

SGS Loulo During the Period

Summary Bivariate Statistics	Gara	Yalea
Correlation	94.13%	98.26%
Slope	0.92	1.07
R2	0.89	0.97
Y Intercept (g/t Au)	0.0325	-0.0212

Figure11-8 and Figure11-9 illustrate a log scatterplot and a duplicate HARD plot of the coarse crush reject duplicates from Gara analysed at SGS Loulo.

18^th September 2018 Page 94

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-8 Log Scatter Plot of Gara Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period

Figure11-9 HARD Plot of Gara Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period

18^th September 2018 Page 95

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-10 and Figure11-11 illustrate a log scatterplot and a duplicate HARD plot of the coarse crush reject duplicates from Yalea analysed at SGS Loulo.

Figure11-10 Log Scatter Plot of Yalea Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period

Figure11-11 HARD Plot of Yalea Course Crush Rejects Duplicate Assayed by SGS Loulo During the Reporting Period

18^th September 2018 Page 96

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

A total of 994 field duplicates samples were analysed for Gounkoto by SGS Loulo during this period. Field Duplicate results from the laboratory show a good correlation between original samples, such that it can be concluded that there is little or no bias in the results (Figure11-12, Figure11-13 and Table11-10). Overall the results returned are acceptable, however they can still be improved with implementation of stringent sampling procedures such as intensifying supervision at all levels and laboratory protocols.

The HARD plot illustrated Figure11-13 in shows good precision between duplicate/original samples assayed. However, this still needs improvement and measures such as sampling supervision and adhering to stringent sampling procedure have been intensified.

Figure11-12 Log Correlational Plot of Gounkoto Field Duplicate Assayed by SGS Loulo

Table11-10 Bivariate Statistics for SGS Loulo Correlational Plot of Gounkoto Field Duplicates

Summary Bivariate Statistics	Value
Correlation	98.73%
Slope	1.01
R2	0.97
Y Intercept (g/t Au)	-0.0018

18^th September 2018 Page 97

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-13 HARD Plot of Gounkoto Field Duplicates Assayed by SGS Loulo

Umpire Assays

Pulp duplicate samples are routinely submitted biannually to an external independent laboratory for umpire analysis. The AMTEL laboratory based in Canada is used as the independent umpire laboratory and samples are submitted biannually from Loulo-Gounkoto site. CRM samples are submitted along with umpire samples to check for bias at the umpire laboratory.

Loulo

A total of 388 pulps duplicates from SGS Loulo were submitted to AMTEL for umpire analysis. This comprises of 370 field samples and 18 CRMs. The combined correlation for all of the field samples was 94.14% (Table11-11).

Sample LU084458 which AMTEL returned as 418 g/t Au and SGS Loulo returned 27.5 g/t Au. This sample has been excluded from the statistics for the purpose of scaling. The bivariate statistics for the Loulo umpire samples is presented in Table11-12. Logarithmic scatterplots of the Yalea, Gara and Baboto Umpire samples are presented in Figure11-14 to Figure11-16.

Table11-11 Loulo Umpire Sample Summary

Dataset	Number of Samples	Correlation (%)
Baboto	125	98.35
Gara	126	94.28
Yalea	119	89.78
All	370	94.14

Table11-12 Bivariate Statistics for Loulo Umpire Assays

Summary Bivariate Statistics	Yalea	Gara	Baboto
Correlation	89.78%	94.28%	98.35%
Slope	1.39	1.13	0.99
R2	0.81	0.89	0.97
Y Intercept (g/t Au)	0.2831	0.3913	0.1071

18^th September 2018 Page 98

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The umpire results returned for Yalea indicate that there is an apparent bias on samples over 8 g/t Au. Two relatively high grade (6.66 g/t Au) OREAS216 CRM samples were submitted alongside the umpire samples with both CRMs returning values at two standard deviations above the certified mean. Results from the 53 OREAS216 CRMs submitted during routine analysis at SGS Loulo returned 98.1% of results within one standard deviation, and 100% within two standard deviations. Additionally, the 46 high-grade (9.25 g/t Au) OREAS208 results from SGS Loulo all returned values within one standard deviation of the mean. This indicates that potentially AMTEL is over-reporting high-grade samples, however there is insufficient high-grade CRMs analysed at AMTEL to validate this. Additional high-grade CRMs will be analysed at AMTEL during 2018 to test their high-grade response. In the QP’s opinion, this high-grade bias does not have a material impact on the assay grades and will not materially impact the Mineral Resource estimate.

The Gara umpire samples appear to show little bias however, a slight apparent bias towards AMTEL could be identified in the high-grade samples over 10 g/t Au, although to a lesser extent than the Yalea umpire samples. In the QP’s opinion, this is not considered as having a material impact on the relative accuracy of the Mineral Resource.

The Baboto umpire samples shows no significant bias.

Figure11-14 Log Scatter Plot of Yalea Umpire Assays

18^th September 2018 Page 99

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure11-15 Log Scatter Plot of Gara Umpire Assays

Figure11-16 Log Scatter Plot of Baboto Umpire Assays

18^th September 2018 Page 100

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

A total of 272 pulp duplicates were sent from SGS Loulo to AMTEL laboratory, Canada for umpire analysis. This compromises of 261 field samples and 11 CRMs. The umpire samples returned an overall correlation of 97.23% (Table11-13). A log scatterplot of the Gounkoto umpire results is presented in Figure11-17.

The umpire results from Gounkoto indicate that there is no significant bias is observed in the SGS Loulo laboratory. CRMs were submitted along with the umpire samples to test the accuracy of the AMTEL laboratory. No issues were observed with the returned grades.

Table11-13 Loulo Umpire Sample Summary

Summary Bivariate Statistics	Gounkoto
Correlation	97.23%
Slope	1.07
R2	0.95
Y Intercept (g/t Au)	0.1896

Figure11-17 Log Scatter Plot of Gounkoto Umpire Assays

11.5 Sample Security

RC samples taken on the rigs are bagged and tied with custom Loulo tags as well as being weighed and documented. RC and diamond core are transported by Randgold personnel from the rig to the core yard or a secure storage facility.

18^th September 2018 Page 101

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Labelled samples are placed into large bags and these are sealed. The samples are placed in a crate which are transported to the SGS Loulo laboratory located within the Mine site. Samples which are to be prepared or analysed outside of SGS Loulo are transported by SGS staff.

All laboratory sample backlogs are actively monitored on a weekly basis and if the backlog becomes excessive, samples are dispatched to SGS Bamako. All samples are contained onsite in a secure sampling facility until they can be dispatched.

Results from all laboratories are emailed to a select group of Project individuals and are later imported into the database by the Database Administrator. A paper certificate is mailed at a later date.

All Project data has been migrated and secured in industry standard Maxwell Geoservices (Maxwell) Datashed SQL database for optimal validation through constraints, library tables, triggers, and stored procedures. All site software application databases will be set up to link back to the main database for information retrieval via an Open Database Connectivity (ODBC) link.

During the migration to the SQL database, initially all assay data was migrated from the Microsoft Access database. Subsequently, all assay data has beenre-imported directly from assay certificates from the laboratory and ranked such that they will have a higher priority than the MS Access imported data.

11.6 Independent Audit

QG completed an audit of the Mineral Resource estimation process, including sample preparation and analysis along with QA/QC. QG observed no issues that required immediate rectification and provided some recommendations of future improvements. QG noted that the laboratory and splitters, with one exception, as clean. An issue with a sample pulveriser that was not left in a clean state between uses was identified. This was immediately rectified and Randgold geologists undertake regular unannounced visits to ensure the procedures are being followed.

QG provided the following QA/QC recommendations all of which have since been implemented by Randgold:

● Too many CRMs (standards) are being used at once. By using a smaller number of standards at any one time, will increase the number of results per standard and give a better feel for performance.

● The recording of dates for CRMs, blanks and duplicates needs standardising to a single format that can be interrogated properly.

● Ideally, a longer period than nine months should be analysed to ensure no trends are occurring.

18^th September 2018 Page 102

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

11.7 Discussion

In the QP’s opinion, the sample preparation is well documented and appropriate for the drilling undertaken. The sample security protocols are good as most samples do not leave the mine site as they are prepared and analysed at SGS Loulo.

The actual insertion rates of QA/QC samples are at or above the insertion rates outlined in Randgold’s SOP. The SOP is deemed to be in line with industry best practices.

The duplicate analysis for both Loulo and Gounkoto has shown there to be little to no bias in the reported results. Blank samples returned no failures at Loulo (100% pass rate), and only one failure at Gounkoto (99.95% pass rate). The blank pass rate can be deemed as good.

The CRM results and the umpire sampling are deemed as good, with no areas of concern being identified.

Overall the QA/QC results returned are acceptable, however they can be improved with implementation of stringent laboratory protocols and procedures, such as a full implementation of a LIMS sample submission and results reporting system to complement the existing LIMS tracking system.

In the QP’s opinion, there are no issues outlined in the sample preparation and analysis, and that the resulting data is suitable for use within a Mineral Resource estimate.

The field duplicate to normal sample ratio will be reviewed with an intention to increase the insertion rate to match that of other QA/QC samples in 2018.

18^th September 2018 Page 103

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

12 Data Verification

Geological logging is completed on hard copy and then entered into the database. Diamond drill hole samples are based on lithological unit and most often vary from 0.8 m to 1.2 m whereas RC down hole sample spacing was conducted at 1 m lengths.

All point samples, drill holes and all geological databases are managed on site and stored in a central SQL database, which stores all forms of geological data including collar, assay, surveys, and other geological variables including lithology, alteration, structural observations, and bulk density data.

The primary Gemcom and Access databases were fully migrated to a Maxwell DataShed SQL database in 2015 by Maxwell Geoservices (Maxwell). The database is administered by the site database administrator and overseen by the Loulo Regional database administrator.

The Maxwell DataShed SQL database is configured for optimal validation through constraints, library tables, triggers, and stored procedures. Data that fails these rules on import are rejected or stored in buffer tables until corrected.

Assay certificates are provided in digital form from the laboratory in the prescribed format. These assay files are imported directly in to the DataShed database. This can only be completed by Database Manager and authorised network users that have undergone Maxwell Datashed Administrator training. The assay files include detailed information about the batch, methods, units, detection limits, and elements assayed. The file also includes all QA/QC data in the sequence of analysis.

The assay data is stored in the SQL database in a normalised format to ensure all required information is stored for each sample, and that multiple assay results can be stored for each sample. Upon import a series of validations takes place to check that the data being imported conforms to the expected and required inputs. This includes detection limit, analysis method, reporting units (ppm, g/t, %) etc. The DataShed assay management system merges all the sampling data into one table and is loaded with the assay result. This system also ensures that the sample numbers cannot be duplicated which reduces the chance of having sample numbermix-ups. Ranking of different assay methods is performed automatically such that only one assay result per element is displayed in the master assay table (tblVWDHAssays). The ranking of different analysis methods in held within a lookup table (tblSYSAssMethod) and any changes to this ranking and must be approved by the onsite Database Manager.

The primary SQL database has an MS Access front end application which links to the main database for data entry, reporting and viewing via ODBC. Other MS Access databases used on site are link to the main database to retrieve information for use in various end user applications software via an ODBC link.

A diagrammatical representation of the DataShed normalised assay structure is shown Figure12-1.

18^th September 2018 Page 104

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure12-1 The Normalised Assay Structure Utilised by DataShed

The use of a normalised assay management system, whereby only authorised and fully trained network users can upload laboratory data from digital files meets industry current best practice.

Daily and weekly backups are made and stored on site. Copies of monthlyback-ups are sent to Bamako and quarterly backups are sent to Johannesburg and London.

12.1 Independent Audits

An independent external database audit was completed by Maxwell in February 2016. Maxwell identified that the majority of resource data within the SQL database was in good order and only minor data issues were identified.

Issues outlined by Maxwell included, assays that were unranked, nine QC records with unmatched drill records, and mismatched drill dates between tables. All data that was flagged as having minor issues was quarantined and corrected before being released from the quarantine table by Randgold.

Continued training and mentoring are ongoing for the database administrators as was recommended by Maxwell.

Following the external audit of the Loulo-Gounkoto database, a JORC (2012) Code compliance certificate was issued by Maxwell.

18^th September 2018 Page 105

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

13 Mineral Processing and Metallurgical Testing

13.1 Summary

The Loulo processing plant uses a carbon in leach (CIL) gold extraction process with a throughput capacity of 4.8 Mtpa. Yalea process plant recovery for the remaining life of mine has been estimated at 84.5% based on historical and current testwork; similarly, Gara is estimated as 94.5%. Yalea recovery is impacted by the presence of arsenic and copper. Arsenic and copper impurities also increase cyanide and oxygen consumption. Gounkoto 2016 Super Pit sampling achieved a recovery of 93%. Gold recovery is maintained above 90% by blending the various ore sources (Yalea / Gara/Other Gounkoto) to control copper and arsenic grades in the mill feed.

The current LOM plan has an average recovery of 92.3%. The average gold recovery in 2017 was 92.7%, an improvement from 2016 (91.0%).

Metallurgy and process plant initial scoping and prefeasibility testwork completed on oxide, transition, and fresh Gounkoto samples indicated that plus 90% recovery would be achievable through gravity followed by cyanide leaching of the gravity tails. Further work completed during the 2010 feasibility study and 2016 Super Pit feasibility study which confirmed these initial results and the predicted recovery has been confirmed based on recoveries achieved in the plant. Extensive work was carried out consisting of detailed comminution, variability run of mine (ROM) leach and gravity recoverability work, and variability leach work on gravity tails. Comminution testwork defined the material to be hard, with high abrasive properties. Cyanide detox testwork, carried out on slurries were successful in significantly reducing cyanide levels. Further testwork was carried out in 2015 and 2016 to ensure representative coverage of the Super Pit ore. Recovery curves have been generated for the expected recoveries based on all testwork to date.

● Oxide LOM recovery is estimated to be 96%.

● Transition: linear relationship with recovery capped at 96% for over 6 g/t Au (1.4041*Au+87.542)

● Fresh: lognormal relationship with recovery capped at 96% for over 25 g/t Au(0.974* ln (Au)+91.56)

● Average LOM recovery of Gounkoto is estimated to be 92.3%.

● Back calculated recoveries for Gounkoto based on the tonnes of ore fed since 2010 also confirm 92.5% recovery has been achieved from Gounkoto ore.

13.2 Gounkoto Testwork

Table13-2 is a summary of the Gounkoto testwork.

A total of 565 samples were testedon-site andoff-site for recovery determination throughout Gounkoto phases up to the 2016 campaign for the expansion of current pit to the Super Pit. These campaigns have concluded that the samples were highly amenable to cyanide leaching

18^th September 2018 Page 106

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

and the most recent Super Pit sampling area achieved recovery results of 92.5% on 478 samples (Table13-1).

Table13-1 Metallurgical Optimum Recoveries by Domain 2016

Domain	Optimum          % Recovery
MZ1	95
MZ2	92
MZ3	92
MZ3	94
HW	93
P64	92

Gounkoto Metallurgical Summation

The following is a summary of the metallurgical testwork:

● Samples collected and tested are spatially representative of the mineralisation mined in the Super Pit.

● Sufficient data deemed available for comminution, hydrometallurgy, and ore characteristics.

● Sampling campaign and testwork program returned results consistent with previous campaigns.

● Sampling and recoveries results from Loulo UG operations, plotted on long-section, demonstrate a full spatial coverage.

● Recovery of plus 90% is achievable on 2016-campaign sampling area.

● LOM overall recovery of 92.5% is recommended based on recoveries obtained to date

● Plant and other infrastructure deemed in suitable condition to achieve the above recovery.

● Operating cost of $18.23/t remains in line with LOM Gounkoto operating cost for 2017.

18^th September 2018 Page 107

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table 13-2 Gounkoto Testwork

Campaign	Objectives	Ore Zone	ID Series	Sample Received	Met Series Composite	Sample Type	Testing & Assays	Metal				Cyanide   Addition (g/t)	SCN   Recovery	CIL   Recovery	Average	Reagents Consumption (kg/t)		Note on     Cyanidation Conditions
								Au (g/t)	As (wt %)	Cu (g/t)	Fe (wt %)				Recovery   (%)	NaCN	Lime
0_2009	Pre- feasibility	MZ1	GK00_series	3	Individual	DDH	Loulo Met Lab					500	95.70	96.90	96.30	NA	NA	Optimal leach conditions
			001…003									5,000	95.10	95.70	95.40			Excess leach conditions
1_2009	Scoping study : On site	MZ1,2,3	Met_CND_series 001..0030	30	LG MG HG	DDH, RC	Loulo Met Lab	12.19				1,000	94.58	95.22	94.90	0.6	1.12	Excess leach conditions
2_2010	Scoping review	Enter deposit Strike Depth	Ox_	10	Composite	Trench RC DDH	SENET SGS KC Africa	8.53	0.02	50.00	5.14	500		92.98	96.00	0.36	0.86	24hr BRT at natural DO
	Pre- feasibility	MZ1,2,3 and HW	Tran_	10	Composite			9.86	0.03	41.00	5.66	500		90.82	94.00	0.36	2.44	24hr BRT at natural DO
	Plant design		10x Sulf_	10	Composite			7.82	0.03	32.91	4.88	500		90.01	91.00	0.34	0.32	24hr BRT at excess DO
5_2010	Feasibility confirmation	Enter deposit Strike_ Depth MZ1,2,3 and HW	Var Samples series 01 .10 (Sulphides)	20	Series 40,000	DDH	SGS SA KC Africa					1,000						Excess leach conditions
6_2010/2011	Petrographic Studies and Met campaign	MZ1,2, and HW	GKMET_PHA SE6series_01 12	12	GKDH series	DDH	Micro- search refer to Lawrence	13.39	0.39	135.92	9.16	500			90.17	0.33	0.71	Optimal leach conditions
												1,000			92.56	0.73	0.62	Excess leach conditions
7_2011	Scoping study : HEAP LEACH	M3	GKMET_PHA SE7_GOUNK OTseries	6	GKDH217 (6x)	Core		12.19
9_2012	Feasibility Polysius AG ; HPGR	MZ1,2,3	Polysius AG	10	GKPQ Met ID not found	DDH	Polysius ?
2014_Camp refer as Campaign 15	Gounkoto OC extension to UG Feasibility study	MZ3_UG and UG_Extension MZ2_UG	GKMet_MZ3D DH 0001	66		DDH Core	Loulo Met Lab	21	2860	95	8.1	400			95.24	0.79	2.35	Optimal leach conditions
			GKMet_MZ3D DH 0002					3.92	540	39.8	4.99	400			92.32	0.63	1.55	Optimal leach conditions
			GKMet_MZ3D DH 0003					2.3	490	22	4.099	400			91.55	0.63	1.47	Optimal leach conditions
			GKMet_MZ2D DH 0001					6.2	0.95	151	8.24	400			91.63	0.83	1.82	Optimal leach conditions
			GKMet_MZ2D DH0002					6.96	0.61	101.4	4.14	400			91.79	0.83	1.71	Optimal leach conditions
			GKMet_MZ2D DH 0003					4.3	0.39	34.5	2.38	400			91.28	0.66	1.35	Optimal leach conditions
			Gounkoto 44651 UG	16			AMTEL 2013	11.62	0.47	550		400			90.00			Optimal leach conditions
				54			ALS 2014					Varied						Variability leach test
2016_Camp	Gounkoto OC extension to a Super Pit Feasibility Study	MZ3	GKMet_MZ3D DH series	40			DDH	4.26				400			88.12	2.14	0.67	Optimal leach conditions
								4.26				1,000			90.02	2.07	1.01	Excess leach conditions
		HW	GS_series	15			DDH	1.99				400			94.15	1.34	0.60	Excess leach conditions
								1.99				1,000			93.87	1.23	0.93	Excess leach conditions

18^th September 2018 Page 108

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

13.3 Sampling and Sample Representativity

Extensive testwork campaigns have been undertaken on all Loulo related orebodies, e.g. Gara and Yalea, and more recently on Gounkoto plus the satellite bodies of P129, P125, Loulo 3, and Baboto since thepre-feasibility stage, from 1985 up until the present time, incorporating both major and minor pits including the underground deeper orebodies. In all cases, special care has been taken to ensure full representivity in terms of:

● Spatial representivity.

● Redox and weathering representivity.

● Lithological representivity.

An example of this is seen for the Gounkoto Super Pit in Figure13-1. This theme is echoed and available for most of the campaigns undertaken to date.

Figure13-1 Gounkoto Cyanidation Sample Locations

13.4 Loulo Metallurgical Testwork

A total of sixteen testwork reports, prior to issue of the first full Loulo feasibility study in 2003, were generated, of which four focused on Heap Leach processing and the balance related to Grind/CIL techniques, pertinent to the adopted process route. Details on the locations of the samples, as well as the nature of the testwork is detailed in the reports. A list of these test reports is presented in Table13-3. The location of the core samples used in the metallurgical testwork is detailed within these reports.

18^th September 2018 Page 109

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table13-3 Summary of Testwork Reports Prior to First Feasibility Study

Report No.	Laboratory	Process Type	Year
84-RDM-067-MIN	BRGM	Grind/CIL	1984
85-DAM-032-MIN	BRGM	Heap Leach	1985
86-MLI-066-MIN	BRGM	Heap Leach	1986
Progress No. 4	BHP	Grind/CIL	1995
ARMC 8481	A R MacPherson	Grind/CIL	1995
MRGE	Microsearch	Mineralogical	1997
S72517	AARL	Grind/CIL	1997
S72653	AARL	Grind/CIL	1997
S72681	AARL	Grind/CIL	1997
37105	OMC/Amtel	Grind/CIL	1997
F1711-2.MPC	Eimco	Grind/CIL	1997
S72706	AARL	Grind/CIL	1997
S72731/1	AARL	Heap Leach	1998
S72731/2	AARL	Heap Leach	1998
MET 01/D81	Lakefield Research	Grind/CIL	2002
(None)	Mintek	Grind/CIL	2002

Subsequent to the feasibility study, there have been numerous other test campaigns focussing on the deeper orebodies pertaining to the underground developments, satellite pits, as well as targeted areas within the larger orebodies, such as the ‘Purple Patch’.

13.5 Recovery

Gold recovery is based on testwork and operational history. Yalea process plant recovery for the remaining life of mine has been estimated at 84.5%, Gara is estimated at 94.4%, Gounkoto is estimated at 92.5 %, and Baboto satellite pits at 94.0%.

Yalea recovery is impacted by the presence of arsenic and copper which reduces recovery by impacting the process of adsorption of the gold onto the CIL tanks. Arsenic and copper impurities also increase cyanide and oxygen consumption. Oxygen consumption is increased due to oxidising the associated sulphides.

All underground diamond drill grade control samples at Yalea are assayed for copper and arsenic. Both copper and arsenic are estimated as deleterious elements into the Mineral Resource model (Figure13-2 and Figure13-3) and are not reported as part of the Mineral Resource inventory. This estimation enables the Loulo mine complex mine plan, and subsequent feed blend, to be optimised to ensure that their concentrations do not reach levels which would cause a reduction in recovery (above 300 ppm Cu and above 4,600 ppm As).

18^th September 2018 Page 110

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 13-2 Yalea Copper Estimation within 2017 Resource Model Looking East

Figure13-3 Yalea Arsenic Estimation within 2017 Resource Model Looking East

The operating philosophy is to maintain process plant gold recovery above 90% by blending the various ore sources (Yalea / Gara/ Gounkoto) to control copper and arsenic grades in the mill feed. This blending process is anticipated to adequately manage copper and arsenic issues over the LOM.

As an analogy of the interaction, both copper and arsenic estimations are included in the block model as a recovery percentage (Figure13-4), using all bottle roll testwork data across the resource. Bottle roll tests are also completed on a composite of every grade control hole drilled as part of the standard underground grade control procedure. The recovery estimation is completed as a single pass ID2 (Inverse Distance squared). For the purpose of the estimation, the primary geological domain of Yalea ‘Purple Patch’ is hard bounded from the rest of the data as it is known to have a lower recovery than the rest of Yalea ore due to the high abundance of arsenopyrite. After a single pass estimate allun-estimated blocks are then coded with the background recoveries of 82.5% for Yalea ‘Purple Patch’ and 84% for Yalea North and South.

18^th September 2018 Page 111

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure13-4 Process Recovery Estimated into Yalea Block Model

The increase in overall recovery (i.e. above pure theoretical prediction) observed in Figure13-5 and Table13-4 can be explained by an improvement in management practices; improved understanding of ore mineralogy, blending and improved oxygen dispersion.

Table13-4 Loulo Processing Plant Overall Gold Recovery in 2016 and 2017 by Month

Month	Yalea   Open Pit   (%)	Gara   Open Pit    (%)	Yalea UG   (%)	Gara UG   (%)	Gounkoto     (%)	Overall     Recovery     (%)
Jan-16	0	0	30	22	48	91.16
Feb-16	0	0	30	25	46	91.61
Mar-16	0	0	30	17	53	90.02
Apr-16	0	0	30	23	47	89.78
May-16	0	0	29	23	48	92.03
Jun-16	0	0	29	19	52	91.12
Jul-16	0	0	35	17	48	91.20
Aug-16	0	0	29	24	46	91.50
Sep-16	0	3	30	24	43	90.25
Oct-16	0	4	24	24	48	90.52
Nov-16	0	14	28	17	41	91.47
Dec-16	0	5	32	19	44	91.50
Jan-17	0	0	29	19	51	92.51
Feb-17	0	0	28	21	52	92.06
Mar-17	0	0	30	24	46	92.65
Apr-17	0	0	29	24	47	92.62
May-17	0	0	27	23	50	92.33
Jun-17	0	0	29	22	49	91.95
Jul-17	0	0	29	22	49	91.66
Aug-17	0	0	35	24	41	91.72
Sep-17	0	2	24	23	50	93.23
Oct-17	0	6	26	21	47	94.35
Nov-17	0	0	28	19	52	93.61
Dec-17	0	0	33	26	41	93.48

18^th September 2018 Page 112

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure13-5 Loulo Processing Plant Overall Recovery in 2016 and 2017

13.6 Deleterious Elements

The initial implementation at Loulo called for an unlined tailings storage facility. During the life of mine, an increase in the arsenic content of the deeper underground ores has been observed. This coupled with the subsequently erected paste plants, which require benign slurry streams, has led to Loulo implementing an enhanced mitigation strategy for the deleterious elements present there.

Arsenic and copper are the deleterious elements of note, and the strategy of mitigation is ore blending to ensure their concentrations do not reach levels that could adversely affect recovery or tailings respectively. It is noted that with the TSF being unlined, it is preferred to keep the supernatant pool as small as possible so as to restrict the effect of a resultant plume having an increased driving force to expand. This has not always been possible to achieve with the effects of wet seasons coupled with plant restrictions to accepting recycled process water.

All of the above issues have been addressed with a formal action plan in place to permanently reduce the size of the supernatant pool, but also to implement a direct mitigation treatment regime of the residue stream prior to it leaving the process plant for the TSF. Thus arsenic, copper, and cyanide tenors are to a large extent eradicated before they have the chance of entering the environment either by controlled discharge or seepage. Environmental borehole monitoring attest to the fact that none of this risk has been manifested to date.

Mitigation occurs inside the plant domain at an area known as the Intermediate Plant orI-Plant, which comprises a series of tanks, thickeners, and ponds used specifically to direct water flows, as needed, for the rest of the mine. This is a technical requirement for the underground paste plants which must use a detoxified coarse tailings as backfill material to fill the stopes. The Intermediate Plant has cyanide destruction together with arsenic fixing andtwo-stage cycloning

18^th September 2018 Page 113

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

to remove clay and fines (if present) and discharges the coarse fraction into a tank. Plant tails also report to this area prior to being pumped to the TSF.

Specific remedial measures include the following:

· Caro’s acid – A proprietary installation sourced from CyPlus used for both the destruction of cyanide species to ensure a (weak acid dissociated (WAD) cyanide tenor of no greater than 50 ppm at the supernatant pond at the TSF. This same Caro’s acid, supplemented as necessary with hydrogen peroxide, acts as an oxidant for a further mechanism described below.

· Ferrous sulphate addition – is oxidised so that the iron is in the ferric state. Arsenic (III) in solution is also oxidised to arsenic (V) state whereupon it links with the iron to form a stable ferric arsenate precipitate which then settles out on the TSF.

· A final note in terms of cyanide remediation is that, although Randgold is not a signatory to the cyanide code, it readily chooses to abide by the principles of the code in all operations.

The above regime has proved successful in reducing the tenors of these deleterious elements, often by more than 80%. Further mitigation lies in the blending of the ore feed to the plant to ensure that inherent or initial arsenic levels remain relatively low in the process plant. Figure13-6 depicts the treatment regime at theI-Plant whereas Figure13-7 shows the effective reduction in tenor.

Figure13-6 Treatment Regime in the Intermediate Plant

18^th September 2018 Page 114

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure13-7 Effective Reduction in Tenor

18^th September 2018 Page 115

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14 Mineral Resource Estimates

The Mineral Resource estimates have been prepared according to the guidelines of the Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI43-101 and there are no material differences.

Definitions for resource categories used in this report are consistent with those defined by CIM (2014) and adopted by NI43-101. In the CIM classification, a Mineral Resource is defined as “a concentration or occurrence of solid material of economic interest in or on the Earth’s crust in such form, grade or quality and quantity that there are reasonable prospects for eventual economic extraction”. Mineral Resources are classified into Measured, Indicated, and Inferred categories. Loulo and Gounkoto Mineral Resources are declared separately.

14.1 Loulo Summary

Geological interpretations and Mineral Resource estimation were completed at Loulo by Randgold with an effective date of 31st December 2017. Table14-1 summarises the Loulo Mineral Resource estimate as of 31st December 2017.

Table14-1 Loulo Gold Mine Mineral Resource Statement as of 31^st December 2017

Source	Mineral Resource	Tonnes (Mt)	Grade (g/t)	Ounces (Moz)	*Attributable Gold (Moz)
Stockpiles	Measured	1.7	1.60	0.086	0.068
Open Pit	Measured	1.9	2.67	0.17	0.13
	Indicated	6.9	3.08	0.69	0.55
	Inferred	2.5	3.3	0.27	0.21
Underground	Measured	17	4.99	2.7	2.1
	Indicated	26	5.18	4.3	3.4
	Inferred	9.8	4.1	1.3	1.0
Total	Measured	20	4.49	2.9	2.3
	Indicated	33	4.73	5.0	4.0
	Measured +     Indicated	53	4.64	7.9	6.3
	Inferred	12	3.9	1.6	1.3

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in Loulo Gold Mine. Mineral Resources are reported on a 100% and attributable basis.

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resource tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are those within a $1,500/oz pit shell at an averagecut-off grade of 0.7 g/t Au.

Underground Mineral Resources are those below the $1,500/oz pit shell at acut-off grade of 1.89 g/t Au at Gara and 2.04 g/t Au at Yalea.

Mineral Resources for Loulo were generated by Mr Simon Bottoms, CGeol, an officer of the company and Qualified Person. Numbers may not add due to rounding.

18^th September 2018 Page 116

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo year end 2017 Measured and Indicated Mineral Resources are estimated at 53 Mt at 4.64 g/t Au containing 6.3 Moz of gold and there is an Inferred Mineral Resource of 12 Mt at 3.9 g/t Au containing 1.6 Moz gold. The Mineral Resources consist of Loulo open pit and underground Mineral Resources and surface stockpiles.

The Loulo Gold Mine resources consist of the Gara and Yalea underground resources and the Baboto, Loulo 3 and Gara West open pit resources. There are six additional minor satellite deposits with previously declared resources, P129, P125L3, P129QT, Loulo 1, Loulo 2 and L2L3Gap, and PQ10. These minor satellite deposits combined make up approximately 1% of the declared Mineral Resources, have not been updated using current practices, are not the target of recent exploration, and do not contain Ore Reserves.

Gara, Yalea and Baboto Mineral Resources have been updated in 2017 by additional drilling and mapping providing further information to improve the geological model for resources estimation. No new data has been added to Loulo 3 and Gara West since the previous model update, and the models remain the same as declared in year end 2016 resources.

Both the Gara and Yalea models incorporate data from the open pit and underground grade control infill drilling as well as exploration diamond drill holes. Thecut-off date for data used to create the models is the 20th October 2017 for Yalea and the 20th November 2017 for Gara. For the Baboto satellite, the datacut-off was 1st December 2017, for Loulo 3 the datacut-off was 12th November 2015, and for Gara West the datacut-off was December 2007. No drilling that would make a material impact on the Mineral Resource was completed between thecut-off date and the effective date of the Mineral Resource (31st December 2017). Open pit and underground scans are completed monthly and annually on 31st December to quantify all mining depletion and the block models are depleted with these during reporting.

All Mineral Resources updates since 2014 at Loulo used Maptek Vulcan. Some minor satellite deposits were created in Gemcom prior to this and have not been updated. Table14-2 summarises the effective model date for the major deposits at Loulo and details if they are currently in production.

Table14-2 Summary of Loulo Deposit and Model Date

Deposit	Producing Status	Model Date
Yalea Underground	Active	31/12/2017
Gara Underground	Active	31/12/2017
Baboto	Unmined	31/12/2017
Loulo 3	Partially Mined	31/12/2017
Gara West	Unmined	31/1/2015
P129	Partially Mined	31/12/2009
P125L3	Unmined	31/12/2011
P129QT	Partially Mined	31/12/2006
Loulo 1	Unmined	31/12/2011
Loulo 2 and L2L3Gap	Unmined	31/12/2011
PQ10	Unmined	31/12/2010

18^th September 2018 Page 117

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Thecut-off grade selected for limiting each of the Mineral Resources corresponds to the insitumarginalcut-off grade using a gold price of $1,500/oz.

For the open pit Mineral Resources, the pit shell selected for limiting each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground panels were reviewed and those that were deemed as having a reasonable prospect of eventual economic extraction were included in the reported Mineral Resource.

In the QP’s opinion there are no environmental, permitting, legal, title, fiscal, socioeconomic, marketing, fiscal, or other relevant factors which could materially impact the Loulo Mineral Resources.

14.2 Gounkoto Summary

Geological interpretations and Mineral Resource estimation were completed at Gounkoto by Randgold with an effective date of 31^st December 2017. Table14-3 summarises the Gounkoto Mineral Resource estimate as of 31^st December 2017.

Table14-3 Gounkoto Gold Mine Mineral Resource Statement as of 31^st December 2017

Source	Mineral Resource	Tonnes (Mt)	Grade (g/t)	Ounces (Moz)	*Attributable Gold (Moz)
Stockpiles	Measured	1.8	1.96	0.11	0.089
Open Pit	Measured	5.4	4.33	0.75	0.60
	Indicated	18	4.04	2.3	1.9
	Inferred	1.4	2.3	0.11	0.08
Underground	Indicated	3.0	5.74	0.56	0.45
	Inferred	2.6	3.51	0.29	0.23
Total	Measured	7.1	3.75	0.86	0.69
	Indicated	21	4.28	2.9	2.3
	Measured + Indicated	28	4.15	3.7	3.0
	Inferred	4.0	3.1	0.40	0.32

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in Gounkoto Gold Mine. Mineral Resources are reported on a 100% and attributable basis.

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resource tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are those within a $1,500/oz pit shell at an averagecut-off grade of 0.8 g/t Au.

Underground Mineral Resources are those below the $1,500/oz pit shell at acut-off grade of 2.0 g/t Au.

Mineral Resources for Gounkoto were generated by Mr Simon Bottoms, CGeol, an officer of the company and Qualified Person.

Numbers may not add due to rounding. Gounkoto year end 2017 Measured and Indicated Mineral Resources are estimated to be 28 Mt at 4.1 g/t Au containing 3.7 Moz gold and Inferred Resources are estimated to be 4 Mt at 3.1 g/t Au containing 0.32 Moz.

Gounkoto Mineral Resources consist of three primary sources: Gounkoto open pit and underground, and Faraba open pit. Gounkoto open pit and underground models were updated in 2017 because of depletion at Gounkoto open pit along with new drilling and model changes for both open pit and underground. Faraba model remains the same as 2016 as no additional work has been undertaken on the deposit.

18^th September 2018 Page 118

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Thecut-off date for data used in the Gounkoto update is 14^th July 2017 and 31^st December 2011 for the Faraba model. No drilling that would make a material impact on the Mineral Resource was completed between thecut-off date and the effective date of the Mineral Resource (31^st December 2017). Open pit scans are completed on 31^st December 2017, to capture all completed mining and the block models are depleted with these during reporting.

All Mineral Resource models for Gounkoto have been created using Maptek Vulcan. Table14-4 summarises the effective model date for each deposit in Loulo and if it is currently in production.

Table14-4 Summary of Gounkoto Deposit and Model Date

Deposit	Producing Status	Model Date
Gounkoto Open Pit	Active	31/12/2017
Gounkoto Underground	Unmined	31/12/3017
Faraba	Unmined	31/12/2015

Thecut-off grade selected for limiting each of the Mineral Resources corresponds to the insitumarginalcut-off grade using a gold price of $1,500/oz.

For the open pit Mineral Resources, the pit shell selected for limiting of each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground panels were reviewed and those that were deemed as having a reasonable prospect of eventual economic extraction were included in the reported Mineral Resource.

In the QP’s opinion there are no environmental, permitting, legal, title, fiscal, socioeconomic, marketing, or other relevant factors which could materially impact the Gounkoto Mineral Resources.

14.3 Resource Database

Loulo

Thecut-off date for data used to create the models is the 20^th October 2017 for Yalea and the 20^th November 2017 for Gara. For the Baboto satellite, the datacut-off was 31^st May 2017, for Loulo 3 the datacut-off was 12^th November 2015, and for Gara West thecut-off was December 2007. No drilling that would make a material impact on the Mineral Resource was completed between thecut-off date and the effective date of the Mineral Resource (31^st December 2017).

The complete resource databases used for the 2017 model updates are detailed in Table14-5. Diamond drill holes (DD) and reverse circulation drill holes (RC) were combined with trench samples to create and estimate the Mineral Resources. Underground channel samples were used at Gara only as due to the vein stockwork nature of this deposit, a cross face channel sample represented a true representation of the mineralisation.

18^th September 2018 Page 119

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-5 Loulo Mineral Resource Dataset

Hole Type	Yalea		Gara		Baboto		Loulo 3		Gara West
	Hole   Count	Meters	Hole   Count	Metres	Hole   Count	Metres	Hole   Count	Metres	Hole   Count	Metres
DDH	1,142	243,720	442	177,264	61	7,873	1,112	1,709	34	8,952
RC	2,229	77,346	2,454	85,875	2,515	93,526	679	24,730	96	7,121
Channel			303	1,710
Trench	331	17,872	21	1,717	95	7,688	4	12,705
Total	3,702	338,937	3,220	266,565	2,671	109,087	1,795	39,144	130	16,073

Gounkoto

Thecut-off date for date used in the Gounkoto update is 14^th July 2017 and 31^st December 2011 for the Faraba model. No drilling that would make a material impact on the Mineral Resource was completed between thecut-off date and the effective date of the Mineral Resource (31^st December 2017).

The complete resource database that was used to report the 2017 Mineral Resource is summarised in Table14-6.

Table14-6 Gounkoto Mineral Resource Dataset

Hole Type	Gounkoto		Faraba
	Hole Count	Metres	Hole Count	Metres
DDH	719	160,791	17	4,385
RC	5,681	293,040	75	8,424
Total	6,400	453,831	92	12,809

14.4 Geological Modelling

Loulo

Yalea

The geology model wireframes were generated in Vulcan on 10 m spaced vertical sections. Wireframe strings are generated on these sections based on trenches, RC, and diamond drill holes, including underground infill grade control drilling, and the hanging wall and footwall shears as bounding structures. There is generally a sharp contact between the mineralisation and waste material with a strong correlation between the grade change and that of the alteration and structure deformation.

The Yalea solid is comprised of Yalea South, Yalea North, P125 Zone, and a high-grade domain referred to as ‘Purple Patch’. These solids were cropped against a basal surface. The basal surface for each deposit was generated at the limit of geological continuity, which is above that of the deepest drill hole intersects. The mineralised model has also been updated to follow the Yalea Shear.

The domain solids have been separated into weathering domains and structural orientation sub-domains (Figure14-1). ‘Purple Patch’ (domain 9006) is located predominantly in the northern portion of the mineralisation and has been domained as a separate high-grade zone due to the

18^th September 2018 Page 120

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

very high sulphide content within this zone which gives it a higher density (average 3.1 g/cm3) This domain is observed in drill core and underground as visually different from the other fresh domains and is observed to have different density and grade characteristics, and therefore can further support it is estimated with a hard boundary.

A completere-log of all the existing drill core from within the ‘Purple Patch’ concluded that ‘Purple Patch’ style mineralisation only occurred in areas where at least two of the three phases of alteration were present (Figure14-2). The application of the thin zone sub domain wireframes within the ‘Purple Patch’ has also enabled the minimum wireframe thickness to be set to 1.5 m and thus minimising the amount of waste samples added to the domain.

Figure14-1 Yalea Model Geological Domains (Looking East). A Portion of 9007 is Also Referred to as Yalea South HG Plunge

During 2017 step out resource definition drilling to the south of the main Yalea resource defined a significant extension to the Yalea South Mineral Resources (9007).

All Yalea South Extension domains remain as unclassified or conceptual within the 2017 resource models as further resource definition and exploration drilling is planned to potentially convert this into a Mineral Resource during 2018.

18^th September 2018 Page 121

18^th September 2018 Page 122

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gara

The Gara geological domain solid model was generated in Vulcan modelling software from 10 m spaced vertical section strings. Wireframe strings were created on these using trenches, advanced RC, grade control RC, and diamond drill holes, including underground drilling, and mapping information. There is generally a sharp shear contact between the mineralisation and hanging wall waste material. The footwall contact is more complex due to the gradational reduction in the degree of tourmalinisation. Due to the folded nature of the body the solids have beensub-divided into separate structural orientation domains (Figure14-3). The mineralisation domains have a strong correlation between the grade and theSi-Ca fracture vein intensity as a function ofsyn-mineralisation deformation.

Continued drilling in 2017 following on from exploration conversion drilling during 2016 defined southerly extensions of the mineralisation. These have been classed under the Gara Far South and Gara Far South Extension and coded as domains 76059600, 4605 and 9605. Domains 4600, 4605 and 8400 are all excluded from underground mineral resource reporting due to being deemed to not have reasonable prospect of eventual economic extraction.

Each geological domain is attributed a single code for the estimation. The homoclinal limb of the fold is coded as geological domains 3000 and 4000 which corresponds to Zone 1 in historic model terminology.

Figure 14-3 Gara Model and Geological Domains

Loulo 3

The mineralisation wireframes and geological solids were generated in Gemcom in January 2012, using a combination of 25 m spaced vertical sections and closer spaced grade control spacing. The interpretation uses trenches, RC and diamond drill holes that intersect the mineralisation. The models are cropped against a basal surface which was constructed at half of the average down dip drill hole spacing below the deepest drill holes.

18^th September 2018 Page 123

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo 3 is separated into three zones: namely MZ2 striking 020°, which is cross cut by subparallel MZ1 NW trending mineralisation and the southern FW zone striking 350°.

A NNE trending,sub-vertical mafic intrusive (5 m to 7 m thick) has intruded along a late fault which cross cuts the mineralisation and separates the southern zone and the SW zones. This has displaced the mineralisation by uplifting the eastern block by between 5 m and 10 m (Figure14-4).

Figure14-4 Loulo 3 Model Geological Domains

18^th September 2018 Page 124

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Baboto

The domain solids were generated in Vulcan using vertical sections with 10 m separations for Baboto South and Baboto Centre. The interpretation used trenches, RC and diamond drill holes that intersect the mineralisation, which dips steeply to the west.

The models are cropped against a basal surface which was constructed at half of the average down dip drill hole spacing below the deepest drill holes. The model was cut by dolerite dykes. Baboto South is separated into a GC and explorationsub-domains, whilst all other zones are split by weathering profiles and cut by overlying transported cover. Figure14-5 illustrate the Baboto domains.

Figure14-5 Baboto Model Geological Domains

Gara West

The domain wireframes and geological solids were generated in Maptek Vulcan during November 2015 using a combination of 20 m spaced vertical sections. The interpretation used RC and diamond drill holes that intersect the mineralisation, which dips steeply to the west and are generallylow-grade (Figure14-6).

The November 2015 update of Gara West added two additional lodes outside of the previous 2012 wireframes. Consequently, the model now contains four individual, parallel mineralised lenses striking NNE and steeply dipping west.

18^th September 2018 Page 125

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure14-6 Gara West Model Geological Domains

Gounkoto

The Gounkoto wireframes were created in Vulcan on 6.25 m and 12.5 m spaced sections depending on the drill spacing. Interpretation strings were created using lithological, alteration, mineralisation, structure measurement logging together with assay data from the surface and pit mapping, the trench, RC, and DD hole data. The strings were connected to form three-dimensional wireframes.

Due to the structural controls that dictate the 3D geometric shape, the alteration type and mineralisation style, Gounkoto has beensub-divided into eight geological domains. There are four main zones (MZ1to MZ4), eleven footwall iron structure NS zones (FWFE), five footwall NE zones (FWNE), four hanging wall zones (HW) and one original satellite (sub divided into P64W and P64E) which is now included into the ‘Super Pit’ (Figure14-7).

18^th September 2018 Page 126

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure14-7 Gounkoto Model Geological Domains

Faraba

The wireframes were generated by drawing 20 m vertical sections perpendicular to the broadlyN-S strike of the deposit in Vulcan geological modelling software. Geologists constructed an interpretation on printouts of the sections, using the drilling, lithological, alteration, mineralisation, and structural measurements. From these interpretations, strings representing the outlines of the domains were digitised (Figure14-8).

18^th September 2018 Page 127

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure14-8 Faraba Model Geological Domains

The main mineralisation is striking generally NS to NNW and dips between 50° and 70°. To the east the mineralisation is truncated by a black sandstone discordance. Faraba mineralisation has beensub-divided into three geological domains:

· Main Zone:Located between two NE faults identified on section 53 to the north and sections 38, 39, 40 and 42 to the south. The faults are characterised by pervasive haematite and strong brecciation.

· Hanging Wall: Weak mineralisation identified mainly in the fine greywacke saprolite material with a weak oxidation from section 36 to 48 (580 m).

· Footwall: Broad mineralisation intersected to the northern part of the structure. Hosted in coarse sediments with weak silica and carbonise alteration with local shearing. Generally, steeply dipping.

18^th September 2018 Page 128

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.5 Topography

Topography has been depicted using a digital terrain model (DTM) LIDAR 5 m resolution surface for all deposits in the Loulo and Gounkoto Permits (clipped from the regional DTM completed by Randgold in October 2010). This topography includes historical artisanal pits.

The topography surface covers the entire Project area as required for mine design purposes. Where surface mining has occurred, updated LIDAR surveys have been completed on a monthly and annual basis. The surface was visually checked against drill hole collar elevations for the Project, and an acceptable match was found.

14.6 Bulk Density

Bulk density measurements were carried out on the mineralised and waste material. Distributions of each group were studied with outliers being excluded before the mean value was calculated for each rock type which was then hard coded into model.

Fresh, transition and saprolite core was obtained from the diamond drilling and the bulk density was calculated using Archimedes principles (water immersion method). Drill core was cut 10 cm for HQ core and 15 cm for NQ core lengths before being measured in air and again in water. Where needed, core was wrapped prior to weighing to prevent water ingress. The mass measurements are accurate to 0.1g and the scale is equipped with a hook and basket for weighing the sample in water. The dry mass and submerged mass of each sample were recorded, and the following formula utilised to calculate the density of the sample:

Density =	mass(air)
	[mass(air) – mass(water)

The scale used to weigh the samples (below) is calibrated using calibration weights from the onsite SGS laboratory.

Density measurements were carried out on the mineralisation and waste material. Distributions of each group were studied with outliers being excluded before mean value was calculated.

Loulo

The assigned density values are based upon the mean or median of the density dataset for each domain/rock type (Table 14-7 to Table 14-10). A single density value is hard coded to the block model for each rock type.

18^th September 2018 Page 129

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table 14-7 Yalea Density Data

Rock Type	SG (g/cm3)
Sulphide	2.79 (9005)
	2.86 (9001,9002, 9003, 9004)
	2.94 (9146)
	9006 –ID2 estimate  & 3.1 background assign
Transition	2.1 (9001, 9003)
	2.47 (9005)
Oxide	1.83
HW Fresh	2.75
HW Transition	2.17
HW Saprolite	2.10
FW Fresh	2.76
FW Transition	2.37
FW Saprolite	2.10
Dolerite Fresh	2.60
Dolerite Transition	2.5
Dolerite Saprolite	2.30

Table 14-8 Gara and Gara West Density Data

Rock Type	SG (g/cm3)
Sulphide	2.82
Transition	2.48
Oxide	2.48
HW Fresh	2.75
HW Transition	2.17
HW Saprolite	1.69
FW Fresh	2.76
FW Transition	2.23
FW Saprolite	1.90
Dolerite Fresh	2.96
Dolerite Transition	2.30
Dolerite Saprolite	2.30

Table 14-9 Loulo 3 Density Data

Rock Type	SG (g/cm3)
Sulphide	2.82 (Main)
	2.77 (South)
	2.75 (SW)
Transition	2.26 (Main)
	2.01 (South)
	2.26 (SW)
Oxide	1.68 (Main)
	1.77
	(South)
	1.68 (SW)
HW Fresh	2.72
HW Transition	2.37
HW Saprolite	2.10
FW Fresh	2.76
FW Transition	2.17
FW Saprolite	2.10
Dolerite Fresh	2.96
Dolerite Transition	2.50
Dolerite Saprolite	2.30
Porphyry Fresh	2.83
Porphyry Transition	2.50
Porphyry Saprolite	2.30

18^th September 2018 Page 130

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table 14-10 Baboto Density Data

Rock Type	SG (g/cm3)
Sulphide	2.72
Transition	2.33
Oxide	1.95
Transported Cover	1.34
Quartzite / SQR Fresh	2.74
Quartzite / SQR Transition	2.34
Quartzite / SQR SAP	1.34
Dolerite Fresh	2.96
Dolerite Transition	2.50
Dolerite Saprolite	2.30

Yalea ‘Purple Patch’ Density Study

Due to the presence of a consistent tonnage bias within the 2017 Loulo mine reconciliation (refer to Section 14.19 Reconciliation) a review of all Yalea density data was undertaken. This confirmed all other Yalea geological domain densities but highlighted significant variability within the ‘Purple Patch’ density distribution (Figure 14-9).

Figure 14-9 Yalea 2017 ‘Purple Patch’ Density Distribution

Previously the ‘Purple Patch’ density had been assigned at 3.1 g/cm³ as the mean of the domain distribution. However, when reviewed spatially it was apparent that the lower ‘fringe’ portions of the ‘Purple Patch’ were a different density to the main portion of the ‘Purple Patch’ (Figure 14-10).

18^th September 2018 Page 131

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 14-10 Yalea 2017 ‘Purple Patch’ Model and Density Data Spatial Distribution Highlighting Lower Domain Contact

The distributions of the separated 9506 and ‘Fringe’ domains show significantly different distributions. The resultant sub domain distributions still show a much larger interquartile range than other domains within Yalea. Consequently, a single pass ID² (Inverse Distance squared) estimate was complete within each of the two sub domains which were hard bounded from one another for estimation.

Different estimation methods were tested and ID² estimation was selected as the appropriate estimation method for density estimation within the 2017 Yalea resource model as this applies slightly less weighting to local outliers relative to ID³ (Inverse distance cubed). After a single pass estimation, the mean density for each sub domain was then assigned to the un-estimated portion. The resultant 2017 density for Yalea Mineral Resource is shown below in Figure 14-11.

Figure 14-11 Yalea 2017 Mineral Resource Block Model Densities

18^th September 2018 Page 132

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The resultant impact of the density estimate was a reduction of 264 kt of rock, but a 5% improvement on the mine call factor ounce reconciliation for 2017 (refer to Section 14.19 Reconciliation)

Gounkoto

A single density was applied for each mineralised domain within the Gounkoto Mineral Resource. As there is no trenching on the Gounkoto deposit, transitional and saprolite material was assigned density values of 2.40 g/cm³ and 1.83 g/cm3 respectively, based on the limited data available and this was verified against other similar Mineral Resources in the Loulo Permit. This represents a low risk to the Mineral Resource estimate because with the exception of P64, all oxide and transition material within the Gounkoto Mineral Resource has been depleted.

The waste domains have been split into hanging wall and footwall sub-domains due to structural dipping and differences in complexity. Outliers are removed from the density values prior to calculations.

Table 14-11 summarises the densities used in the Gounkoto 2017 Resource update.

Table 14-11 Densities Applied Gounkoto for 2017 Resources

Material	Unit	MZ1	MZ2NW, MZ2NW, MZ2NWFW	MZ3	MZ4, MZ4FW	HW1 to 4, MZ3HW	FWFE1 to 11, FWNE1 to 5	P64W, P64E	HW Waste	FW Waste
Sap	g/cm3	1.83	1.83	1.83	1.83	1.83	1.83	1.83	1.68	1.86
Trans	g/cm3	2.4	2.4	2.4	2.4	2.4	2.4	2.4	2.4	2.56
Fresh	g/cm3	2.81	2.98	2.86	2.81	2.79	2.81	2.81	2.74	2.76

MZ 2 has a significantly higher fresh rock density than the other zones due to a significant increase in the iron carbonate and haematite alteration.

Faraba

At Faraba, single density values were applied to each lithology sub split into weathering sub domains as outlined in Table 14-12. Any lithologies that had insufficient samples to ascertain a density used the equivalent density obtained for Gounkoto.

Table 14-12 Densities Applied Faraba for 2017 Resources

Material		Unit	Sandstone (100)	Greywacke (200)	SQR (300)	QR (400)	Limestone (500)	Intrusive (600)
Ore	Saprolite	g/cm3	1.89	1.88	2.35	2.35	1.36	2.30
	Trans	g/cm3	2.57	2.36	2.50	2.59	1.91	2.50
	Fresh	g/cm3	2.84	2.72	2.78	2.78	2.89	2.76
Waste	Saprolite	g/cm3	1.88	1.88	2.35	2.35	1.36	2.30
	Trans	g/cm3	2.57	2.36	2.50	2.59	1.91	2.50
	Fresh	g/cm3	2.72	2.72	2.78	2.78	2.77	2.78

18^th September 2018 Page 133

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.7  Compositing

Drill hole data is composited prior to top cutting. The aim of compositing drill holes is to eliminate grade bias due to sample length during interpolation. Prior to selection of the final composite length, the data is visually analysed using a histogram of sample length to identify the mode of the length. The coefficient of variation (CV), standard deviation (SD) and mean plots were produced with several composite lengths to ensure that the mean, CV, and SD are stable and do not increase with composting.

All drill hole information within the mineralised zones is composited, with a minimum sample length applied so that any sample below this is removed and thus not used during estimation. This also allows compositing of up to the composite length + minimum sample length (e.g. 2.0 m + 0.5 m = 2.5 m sample length) at the edge of mineralisation. The non-composited residuals (those below minimum sample length) are analysed for each estimation domain to ensure that the discarded points do not bias the remaining data populations.

The composite samples are flagged using the mineralised domains. Missing sample intervals are ignored during compositing, although the sources of missing samples are investigated to ensure that there is no bias.

Loulo

Yalea

Data within the estimation domains has been composited to 2.0 m using Vulcan. The compositing method applied attempted to create a maximum sample composite length of 3.0 m and a minimum length of 1.0 m. Most composite lengths are 2.0 m.

Drill holes samples with missing values of -999 have been omitted from the compositing. Composites with length 1.0 m or less are reset to an absent value, in Vulcan this is set to -999 and ignored in the estimate, a total of 54 samples with a maximum grade of 16.5 g/t Au and an average grade of 2.75 g/t Au were impacted. The resultant composite database samples for the underground area are presented in Table 14-13.

Table 14-13 Yalea Total Model 2 m Composite Dataset

Domain	9001	9002	9003	9004	9005	9006	9007
Number of Points	625	408	2,279	2,387	820	1,659	69
Min Au Grade (g/t)	0.01	0.01	0.01	0.01	0.01	0.09	0.01
Uncapped Max Au Grade (g/t)	53.65	44.40	40.56	124.75	117.85	116.65	98.80
Uncapped Mean Au Grade (g/t)	3.10	5.70	3.27	4.19	2.98	9.32	9.35
CV	1.55	1.08	1.09	1.78	2.22	0.93	1.64

18^th September 2018 Page 134

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gara

Data within the estimation domains has been composited to 2.0 m using Vulcan. The compositing method applied attempted to create a maximum sample composite length of 3.0 m and a minimum length of 1.0 m. Most composite lengths are 2.0 m.

Residuals of less than 1.0 m in length are reset to an absent value, in Vulcan this is set to -999 and ignored in the estimate, a total of 82 samples with a maximum grade of 32.74 g/t Au and an average grade of 2.59 g/t Au were impacted. No hard boundaries are applied to the Gara estimation. The resultant composite database samples for the entire underground area is presented in Table 14-14:

Table 14-14 Gara Total Model 2 m Composite Dataset

Domain	9001
Number of Points	8,820
Min Au Grade (g/t)	0.01
Max Au Grade (uncapped g/t)	322.10
Mean Au Grade (uncapped g/t)	4.13
CV	1.85

Loulo 3

A 1.0 m composite is applied with a minimum sample length of 0.5 m. This creates a maximum sample composite length of 1.5 m and a minimum length of 0.5 m. Over 95% of the composites are at 1.0 m length. This method minimises the impact of residuals and so any remaining residuals of less than 0.5 m in length are reset to an absent value and these samples are subsequently ignored (treated as non-existent). For Loulo 3 estimation dataset, a total of with 338 residuals less than 0.5 m in length had values reset to -999. The resultant composite database samples are presented in Table 14-15.

Table 14-15 Loulo 3 Total Model 1 m Composite Dataset

Domain	MZ 1000	SZ 2000	SWZ 3000	Total
Number of Points	6,036	2,822	407	9,265
Min Au Grade (g/t)	0.01	0.01	0.01	0.01
Uncapped Max Au Grade (g/t)	138.87	116.53	98.9	138.87
Uncapped Mean Au Grade (g/t)	2.77	4.02	3.24	3.13
CV	2.31	2.49	2.28	2.43

Baboto

A 1.0 m composite is applied with a 0.5 minimum sample length. This creates a maximum sample composite length of 1.5 m and a minimum length of 0.5 m but retains >95% of the composites at 1.0 m length. This method minimises the impact of residuals and so any remaining residuals of less than 0.5 m in length are reset to an absent value and these samples are subsequently ignored (treated as non-existent). The resultant composite database samples are tabulated in Table 14-16.

18^th September 2018 Page 135

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table 14-16 Baboto South and Centre 1 m Composite Dataset

Domain	1001	1002	1003	1004	2001
Number of Points	24,871	1,938	2,202	106	3,325
Min Au Grade (g/t)	0.005	0.005	0.005	0.005	0.005
Uncapped Max Au Grade (g/t)	45	45	45	31	11.56
Uncapped Mean Au Grade (g/t)	1.97	1.53	1.71	1.36	0.97
CV	1.77	1.82	2.31	2.79	1.73

Gara West

A 1.0 m composite is applied with a 0.5 minimum sample length. This creates a maximum sample composite length of 1.5 m and a minimum length of 0.5 m but retains >95% of the composites at 1.0 m length. This method minimises the impact of residuals and so any remaining residuals of less than 0.5 m in length are reset to an absent value and these samples are subsequently ignored (treated as non-existent). The resultant composite database samples are tabulated in Table 14-17.

Table 14-17 Gara West Total Model 1 m Composite Dataset

Domain	1000	2000	3000	4000	Total
Number of Points	57	552	504	385	1,498
Min Au Grade (g/t)	0.04	0.01	0.01	0.01	0.01
Uncapped Max Au Grade (g/t)	6.67	83.33	38.88	23.60	83.33
Uncapped Mean Au Grade (g/t)	1.37	3.01	2.18	1.84	2.37
CV	1.03	2.41	1.61	1.73	2.18

Gounkoto

Gounkoto

A 2.0 m composite with a minimum 1.0 m composite length was applied to the Gounkoto data. This resulted in a maximum composite sample length of 3.0 m and a minimum of 1.5 m. Table 14-18 outlines the data checks undertaken to ensure that the compositing process does not result in grade bias. The contained metal for the total raw interval data and composite data match closely.

Composite samples below one metre in length are removed from the composite table. Residual analysis was undertaken to ensure that this did not significantly impact the grades. Overall the proportion of the residuals sample within the domains is 4% which is deemed acceptable.

18^th September 2018 Page 136

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table 14-18 Gounkoto 2 m Compositing Data

Domain	Raw Data		2 m Composite Data		% Difference
	Total Interval (m)	Length x Grade (gm)	Total Interval (m)	Length x Grade (gm)	Total Interval	Metal Content
1001	53,888	253,235	53,887	253,233	0	-2
1002	8,419	54,406	8,419	54,379	0	-27
10022	104	555	104	555	0	0
10023	40	87	40	87	0	0
1003	13,402	55,468	13,399	55,467	-3	-1
10032	397	366	397	366	0	0
1004	2,671	8,762	2,671	8,762	0	0
10042	93	141	93	141	0	0
1005	2,946	4,162	2,944	4,157	-2	-5
10051	3,919	5,506	3,919	5,506	0	0
10052	404	262	404	262	0	0
1006	3,589	8,590	3,589	8,590	0	0
1007	1,101	1,720	1,101	1,720	0	0
1008	16	37	16	37	0	0
1009	6	15	6	15	0	0
2001	13,636	19,932	13,636	19,932	0	0
2002	166	250	166	250	0	0
2003	391	328	391	328	0	0
2004	246	532	246	532	0	0
3001	1,688	2,990	1,688	2,990	0	0
3002	288	377	288	377	0	0
3003	5,080	14,120	5,079	14,120	-1	0
3004	1,699	5,015	1,697	5,015	-2	0
3005	245	282	245	282	0	0
3006	24	47	24	47	0	0
3007	169	266	169	266	0	0
3008	35	120	35	120	0	0
3009	34	44	34	44	0	0
3010	33	27	33	27	0	0
3011	65	132	65	132	0	0
4001	68	63	68	63	0	0
4002	50	64	50	64	0	0
4003	95	144	95	144	0	0
4004	134	191	134	191	0	0
4005	145	166	145	166	0	0
Waste	260,532	42,402	260,540	42,439	8	37
Total	375,814	480,805	375,814	480,808	0	2

Faraba

At Faraba 1.0 m composites were chosen as this is the same as the raw sample interval (RC drilling), therefore more than 80% of the samples already have a 1.0 m length. When compositing to 2.0 m, more than half of the data points were reduced. Table 14-19 illustrates differences in samples and mean grade.

Table 14-19 Faraba Composite Sample Analysis

Description	Raw	1 m Composites	2 m Composites	1 m Mean Diff (%)	2 m Mean Diff (%)
Count	2658	2545	1303	-4%	-51%
Length Weighted Mean	1.54	1.54	1.54	0%	0%
CV	2.19	2.05	1.65	-6%	-25%

18^th September 2018 Page 137

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.8   Treatment of High Grades (Top Cutting)

Multifactor data analysis was completed on each separate geological domain with the aim of identifying any grade outliers and choose an optimum top cap value for each of the orientation domains. The multifactor analysis tools applied to this are:

· Histogram plot.

· Log probability plot.

· Disintegration analysis.

· Mean and CV curve to look for the stability point.

· Metal at Risk.

Loulo

Yalea

The top cap analysis results for Yalea are in Table 14-20. Domains 9002 and 9003 were not top capped as no outliers were identified. Another constraint was put in place to limit high value outliers with the ‘Purple Patch’ (domain 9006) which was enforced during the estimation process based on the next significant disintegration. An example of the histograms and log probability plots for Yalea domain 9001 is presented in Figure 14-12.

Table 14-20 Yalea Top Cutting Values Applied to Composites

Data Source	Domain	Top Cap (g/t)	Raw Mean Grade (g/t)	Capped Mean Grade (g/t)	Mean Decrease	Raw CV	Capped CV	CV Decrease
Yalea Underground	9001	70	4.61	4.59	0.43%	1.46	1.36	6.85%
	9002	No Top Cap Applied
	9003	No Top Cap Applied
	9004	65	4.10	4.00	2.44%	1.81	1.57	13.26%
	9005	80	3.60	3.58	0.56%	1.95	1.85	5.13%
	9006	60	9.44	9.37	0.74%	0.93	0.88	5.38%

18^th September 2018 Page 138

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 14-12 Yalea 9001 Grade Distribution (Left: Log Histogram & Right: Log Probability)

Gara

A high-grade top cap has been applied to all geological domains at 70.18 g/t Au with the exception of some of the domains mentioned below in Table 14-21. This transforms the distribution to a lognormal distribution with a good skewness and CV appropriate for ordinary kriging.

Table 14-21 Gara Top Cutting Values Applied to Composites

Data Source	Domain	Top      Cap (g/t)	Raw Mean     Grade     (g/t)	Capped     Mean Grade (g/t)	Mean   Decrease	Raw   CV	Capped   CV	CV   Decrease
Gara Underground	3000 4000 5000 6000 7000 8000 8400 8900 9000 9400 10000	70.18	3.75	3.72	0.99%	1.76	1.54	12.53%
Gara Far South and Gara Far South Extension	9600 9605 4600 4605	23.00	5.15	4.66	9.47%	1.57	1.12	28.67%
Gara OP	7400 7600 7605	20.00	2.85	2.71	4.85%	1.58	1.37	13.32%

Loulo 3

Loulo 3 contains 3 discrete domains; the main zone striking 020°, which is cross cut by two subparallel pods of mineralisation on a left-hand jog which are the southern zone and the SW zone. In addition, a NNE trending, sub-vertical mafic intrusive (5 m to 7 m thick) has intruded along a late fault which separates and offsets the southern zone and the SW zones.

Due to the cross-cutting relationships and the implications on relative timing the grade distribution of each domain is assessed independently to identify if any high-grade outliers that are not spatially correlated exist and as such top caps are applied as detailed in Table 14-22.

18^th September 2018 Page 139

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table 14-22 Loulo Top Cutting Values Applied to Composites

Data Source	Domain	Top     Cap (g/t)	Raw Mean     Grade (g/t)	Capped   Mean Grade   (g/t)	Mean   Decrease	Raw     CV	Capped     CV	CV     Decrease
Main Zone	1000	64.00	2.77	2.74	1.08%	2.31	2.22	3.90%
South Zone	2000	75.19	4.02	3.94	1.99%	2.49	2.36	5.22%
South West Zone	3000	45.50	3.24	3.14	3.09%	2.28	2.07	9.21%

Baboto

Baboto contains two geographically separate zones: Baboto South, and Baboto Centre. The Baboto dataset is heavily clustered in the Centre areas due to the influence of grade control drilling, however, the Baboto South the declustering is negligible. Consequently, a cell declustering of 15 m by 20 m by 15 m is applied to the Baboto Centre estimation domains and combined with top cutting, results in a decrease in mean grade.

Due to the cross-cutting relationships and the implications on relative timing the grade distribution of each domain is assessed independently to identify if any high-grade outliers that are not spatially correlated exist and as such top caps are applied as detailed in the Table 14-23.

Table 14-23 Baboto Top Cutting Values Applied to Composites

Data Source	Domain	Top Cap       (g/t)	Raw Mean   Grade   (g/t)	Capped Mean   Grade (g/t)	Mean   Decrease	Raw     CV	Capped     CV	CV   Decrease
Baboto South	1001	45.00	2.02	1.97	2.48%	2.39	1.77	25.94%
	1002	45.00	1.58	1.53	3.16%	2.40	1.82	24.17%
	1003	45.00	2.00	1.71	14.50%	5.57	2.31	58.53%
	1004	45.00	1.99	1.99	0.00%	2.711	2.711	0.00%
Baboto Central	2001	11.56	1.05	0.97	7.62%	2.29	1.73	24.45%

Gara West

Gara West contains four discrete lodes all striking 020° and west dipping (50° to 70°) with small to moderate changes of direction and dip. However, there is not a clear folding within the lodes. Due to the physical separation, for each of the lodes the grade distribution is assessed independently. The top caps are applied as detailed in Table 14-24.

Table 14-24 Gara West Top Cutting Values Applied to Composites

Data   Source	Domain	Top Cap       (g/t)	Raw Mean   Grade (g/t)	Capped   Mean Grade   (g/t)	Mean     Decrease	Raw     CV	Capped     CV	CV   Decrease
Gara West	10000	45.00	1.60	1.60	0.00%	1.09	1.09	0.00%
	20000	45.00	3.00	2.81	6.33%	2.31	1.87	19.05%
	30000	45.00	2.07	2.05	0.97%	1.63	1.55	4.91%
	40000	45.00	1.96	1.94	1.02%	1.64	1.6	2.44%

18^th September 2018 Page 140

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

Due to the structural complexity which controls the 3D geometry and the data distribution, the Gounkoto deposit has been split into 47 orientation domains within the mineralisation wireframe, including one waste domain which is not constrained by a boundary.

In total 19 estimation domains were top-cut at values between 1.73 g/t Au and 87.5 g/t Au, while a high restriction has been applied in 28 estimation domains (Table 14-25).

Table 14-25 Gounkoto Top Cutting

Block Model Domain Code	Estimation Domain Code	Estimation Domain Name	Top Cut Grade (g/t Au)	HG Constraint (g/t Au)
1001	1	MZ1 East dipping (domain1)	-	3.42
	2	MZ1 East dipping (domain2)	-	96.43
	6	MZ1 Central west dipping (domain6)	-	129.00
	9	MZ1 Central west dipping (domain9)	-	5.00
	3	MZ1 west dipping (domain3)	-	70.00
	4	MZ1 East dipping (domain4)	-	70.00
	5	MZ1 East dipping (domain5)	-	37.35
1002	-	MZ2 NW OP (above -80RL)	-	59.43
		MZ2 NW UG (below -80RL)	-	42.00
10022	-	MZ2 NE	-	32.00
10023	-	MZ2NW FW	-	10.11
1003	7	MZ3 NNW OP	-	79.00
	7	MZ3 NNW UG	-	58.84
	8	MZ3 NNE OP		25.00
	8	MZ3 NNE UG		6.66
	13	MZ3 High-Grade Zone	87.50	-
10032	-	MZ3 HW	-	-
1004	-	MZ4	-	43.00
10042	-	MZ4 FW	-	3.00
1005	-	P64W1	-	31.00
1006	-	P64W2	-	25.00
1007	-	P64W3	-	14.00
1008	-	P64WHW1	2.00	-
1009	-	P64WHW2	1.73	-
10051	10	P64E1 south	-	13.00
	11	P64E1 north	-	17.00
10052	12	P64E2	-	2.40
2001	-	HW1	-	26.00
2002	-	HW2	-	6.50
2003	-	HW3		4.65
2004	-	HW4		8.00
3001	-	FWFE1	24.00	-
3002	-	FWFE2	6.00	-
3003	-	FWFE3	36.00	-
3004	-	FWFE4	44.00	-
3005	-	FWFE5	4.00	-
3006	-	FWFE6	2.70	-
3007	-	FWFE7	10.50	-
3008	-	FWFE8	9.00	-
3009	-	FWFE9	3.00	-
3010	-	FWFE10	1.43	-
3011	-	FWFE11	5.00	-
4001	-	FWNE1	3.00	-
4002	-	FWNE2	5.00	-
4003	-	FWNE3	6.50	-

18^th September 2018 Page 141

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Block Model     Domain     Code	Estimation Domain Code	Estimation Domain Name	Top Cut Grade (g/t Au)	HG Constraint (g/t Au)
4004	-	FWNE4	9.00	-
4005	-	FWNE5	4.50	-
9999 (Waste)	-	Waste (HW/FW)	-	8

Faraba

The top cutting applied to the 2016 Faraba resource is presented in Table 14-26. This shows that the three zones were capped at values of 8.65 g/t Au, 13.36 g/t Au, and 20.79 g/t Au.

Table 14-26 Top Cutting Values Applied to Composites

Data Source	Domain	Top Cap (g/t)	Raw Mean Grade (g/t)	Capped Mean Grade (g/t)	Mean Decrease	Raw CV	Capped CV	CV Decrease
Faraba	MZ (1000)	20.79	1.67	1.61	3.59%	2.02	1.72	14.85%
	HW (2000)	13.36	0.80	0.76	5.00%	2.63	2.31	12.17%
	FW (3000)	8.65	0.86	0.81	5.81%	2.03	1.59	21.67%

14.9 Variography

Exploratory data analysis (EDA) and variography was conducted by Randgold using Snowden Supervisor v8 statistical software on declustered data. The declustered composites are used for spatial analysis and as well to compare with block grade, during the validation process.

Variography was used to analyse the spatial continuity and relation within each of the main domains to determine appropriate search strategies and estimation. The variogram modelling process followed involved the following steps:

· A normal score transform was applied to all data prior to undertaking variography on the two metre, top capped, declustered composite dataset;

· Calculate and model the omni-directional or down hole variogram to characterise the nugget effect;

· Systematically calculate orientated variogram in three dimensions to identify the plane of greatest continuity;

· Calculate a variogram fan within the plane of greatest continuity to identify the direction of maximum continuity within this plane.

· Model experimental variogram in the direction of maximum continuity and the orthogonal directions.

· Apply a back transform to all variogram models to obtain the appropriate variogram models for interpolation of raw composite data.

Where an individual domain has insufficient samples to undertake variography, the variography parameters from a comparative domain with a similar trend was used and the orientation adjusted to match the domain with insufficient data.

18^th September 2018 Page 142

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Variogram Validation

Prior to interpolation runs, each semi-variogram model is cross validated to ensure that any bias in estimated grades compared to the actual sample grades are minimal. This was checked by estimating a grade value at each composite sample point, which ignored said sample point. The resulting grade is compared to the actual sample grade in the same location and is plotted on a scatter plot to establish a possible trend or bias and relative standard error. In most cases, there is level of smoothing in an estimated grade compared to the actual sample grade, but overall, estimated grades and sample grades match well and conditional bias is minimal.

Loulo

Yalea

The total Yalea solid is comprised of seven separated solids that are split into the South, North, P125 Zones, and a high-grade domain referred to as the ‘Purple Patch’ (PP). Cell declustering was completed on the data for some domains prior to the variogram modelling.

Within all the domains from Yalea relative nugget effects ranged between 15% and 25% indicating a relatively low to moderate grade variability. There has been little change from the previous model, except for a small plunge change in Yalea South where it was previously flat. Figure 14-13 illustrates an example of the Yalea South variogram model and back transformed model.

Figure 14-13 Yalea South Variogram Including 9001 and 9002 Domains

18^th September 2018 Page 143

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gara

Gara does not contain any discrete domains which are separated due to differences in the geological characteristics or due to the grade distribution. The only domain splits that are applied to Gara are to account for data density and different physical orientations (changes in dip and strike). Consequently, all data density and orientation domains use soft boundaries for estimation purposes. Cell declustering was completed on the data for some domains prior to the variogram modelling.

Due to the complexity of the folding and subsequent impact on the spatial continuity, the variogram model is based exclusively on data from the orientation sub-domain 3000, which is the largest single monoclonal limb domains and contains over 50% of the total data.

The 3000 domain variogram was modelled (Figure 14-14), and the results re-orientated and applied to the rest of the geological domains.

Figure 14-14 Gara Variogram for the 3000 Domain which was Applied to All Domains with the Bearing, Plunge and Dip Changing Based on Domain Orientation

The nugget for Gara is 36% which is typical for a veining and complex folding deposit such as Gara.

18^th September 2018 Page 144

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo 3

Due to the cross-cutting relationships and the implications on relative timing, individual variogram models are derived for each of the geological domains. Cell declustering was completed on the data for some domains prior to the variogram modelling. The Main Zone and South Zone in Loulo 3 have nugget effects ranging between 11% and 26%, indicating a relatively low to moderate of grade variability. An example of MZ2 domain variogram model and back transformed model are presented in Figure 14-15 and Figure 14-16.

The FW Zone variogram has quite a high nugget effect of 35% which in conjunction with a relatively high CV indicates that there are potentially mixed populations in the grade distribution. Therefore, further geological sub-domaining may be required within future updates if justified by geological modelling or by added data.

Figure 14-15 Loulo 3 Main Zone 2 Individual Structure Normal Score Variogram Models

18^th September 2018 Page 145

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 14-16 Loulo 3 Main Zone 2 Nested Back Transformed Variogram Model

Baboto

Due to the cross-cutting relationships and the implications on relative timing the grade distribution individual variogram models are derived for Baboto Centre and Baboto South. Cell declustering was completed on the data for some domains prior to the variogram modelling. Baboto Centre has a nugget effects ranging of 17% indicating a relatively low to moderate of grade variability. Figure 14-17 and Figure 14-18 illustrate the variogram model and back transformed model for Baboto South.

18^th September 2018 Page 146

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 14-17 Baboto South Variogram

Figure 14-18 Baboto South 2017 Nested Back Transformed Variogram Model

18^th September 2018 Page 147

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gara West

The total combined data variogram continuity range is around 37 m where as for the other individual domains the maximum continuity is above 50 m, indicating that for Gara West individual variograms per lode give better continuity to the estimate instead of mixing all data. Cell declustering was completed on the data for some domains prior to the variogram modelling.

Consequently, individual variogram models are derived for domains 20000, 30000 and 40000. However, due to the low number of samples within domain 10000, robust variogram models could not be constructed. Consequently, the variography from all data combined is applied to domain 10000.

An example of the Gara West domain 20000 variogram model and back transformed model are presented in Figure 14-19 and Figure 14-20.

Figure 14-19 Gara West Domain 20000 Individual Structure Normal Score Variogram Models

18^th September 2018 Page 148

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 14-20 Gara West Domain 20000 Nested Back Transformed Variogram Model

Gounkoto

Nine orientation domains (MZ1 east dipping, MZ1 west dipping, MZ1 central west dipping, MZ2NW, MZ3, MZ4, P64W1, P64W2 and HW1) were selected for variography studies. Cell declustering was completed on the data for some domains prior to the variogram modelling.

Where some domains lacked sufficient data points to support variogram modelling, variography parameters of similar (geologically and spatially) domains were applied. Due to the positively skewed data distribution, a normal score variogram has been chosen to reduce the impact of the extreme outliers and to have a robust back-transformation for estimation.

The directional variogram has been calculated using a normal score transformation on the data. An example of the analysis of the gold grades modelled semi-variograms and resulting estimation input parameters for MZ1 east dipping are summarised in Figure 14-21 and Figure 14-22.

18^th September 2018 Page 149

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure 14-21 Modelled Semi-Variograms for Gounkoto MZ1 Domain 1000 (East Dipping)

Figure 14-22 Gounkoto MZ1 Domain 1000 Nested Back Transformed Variogram Model

18^th September 2018 Page 150

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Faraba

At Faraba, as at Gounkoto, the data is positively skewed, therefore a normal score variogram has been chosen to reduce the impact of the extreme outliers and to have a robust back-transformation for estimation. An example of the Faraba domain 1000 variogram model and back transformed model are presented in (Figure 14-23 and Figure 14-24).

Figure 14-23 Faraba Domain 1000 Modelled Semi-Variograms

Figure 14-24 Faraba Domain 1000 Nested Back Transformed Variogram Model

18^th September 2018 Page 151

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.10 Block Model Estimation and Quantitative Kriging Neighbourhood Analysis

All models are sub-domained by data density such that shorter search ranges could be achieved within the grade control sub-domains. Thus, smaller more localised searches could be applied to the estimation of the grade control domains relative to wider space exploration domains.

When selecting the appropriate block size consideration is given for selectivity during mine design and planning relative to the geology, spatial variability and drill spacing. The Mineral Resources are generated with the multiple block size using different sub sizes.

Each estimation domain has been attributed its own estimation parameters defined from a set of quantitative kriging neighbourhood analysis (QKNA). The QKNA is utilised to optimise the block size, search ranges, sample numbers and discretisation. Optimisations look at Kriging Efficiency (KE) and Slope of Regression (SR) to minimise negative kriging weights.

Figure 14-25 illustrates an example of the QKNA analysis results for Gounkoto MZ1 East Dipping Grade Control Area. The quantitative kriging neighbourhood analysis (QKNA) has been completed in each variogram domain with the first pass of estimation.

Figure 14-25 QKNA Results for Gounkoto MZ1 East Dipping Grade Control Area

18^th September 2018 Page 152

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo

Yalea

In Yalea there are three different areas with different drill hole density. There is a grade control area (average drilling spacing 10 m to 30 m), an advanced grade control area (average drilling spacing 30 m to 80 m) and an exploration area (greater than 80 m). The block size was selected considering the size and shape of the mineralisation, the mining selectivity, the spacing of the drilling grid, and an optimisation exercise using each variogram domain.

The same exercise was applied to optimise the sample number where the minimum and maximum samples were selected based on the drilling spacing in the domain and the stability of the result (KE and SR).

Table 14-27 summarises the parameters from QKNA for Yalea grade control, advance grade control and exploration domains. The block model name for Yalea is “20171020_yalea_uggc_model.bmf” and is a sub cell model generated from Vulcan.

18^th September 2018 Page 153

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-27 QKNA Parameters from Yalea

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
9103	5	15	10	1	45	21	6	8	16	3	5	4
				2	90	42	12	6	14
				3	180	84	24	4	10
				4	360	168	48	4	10
9104	5	15	10	1	45	21	6	8	16	3	5	4
				2	90	42	12	6	14
				3	180	84	24	4	12
				4	360	168	48	4	10
9204	8	24	16	1	90	42	12	12	24	3	5	4
				2	180	84	24	10	20
				3	360	168	48	8	18
				4	720	336	96	8	16
9146	5	15	10	1	45	21	6	6	18	3	5	4
				2	90	42	12	4	16
				3	180	84	24	4	14
				4	360	168	48	4	14
9101	5	15	10	1	74	20	7	8	16	3	5	4
				2	147	40	14	6	14
				3	294	80	28	4	12
9201	8	24	16	1	147	40	14	8	16	3	5	4
				2	294	80	28	6	14
				3	588	160	56	4	12
				4	117	320	112	4	10
9301	10	30	20	1	147	40	14	8	16	3	5	4
				2	294	80	28	6	14
				3	588	160	56	4	10
				4	117	320	112	4	10
9102	5	15	10	1	74	20	7	8	16	3	5	4
				2	147	40	14	6	14
				3	294	80	28	4	12
9202	8	24	16	1	147	40	14	8	16	3	5	4
				2	294	80	28	6	14
				3	588	160	56	4	12
9302	10	30	20	1	147	40	14	8	20	3	5	4
				2	294	80	28	6	18
				3	588	160	56	6	16
				4	117	320	112	4	14
9005	5	15	10	1	37	9	6	8	18	3	5	4
				2	56	14	9	8	16
				3	83	20	14	6	14
				4	125	30	20	4	12
				5	187	46	30	4	10
				6	281	68	46	3	8
9106 & 9156	5	15	10	1	45	21	6	8	18	3	5	4
				2	90	42	12	6	16
				3	180	84	24	4	14
9206	8	24	16	1	75	34	9	12	24	3	5	4
				2	150	68	18	10	20
				3	300	136	36	8	18

Gara

After the geological domaining, estimations domains have been defined and parameters optimised for each domain and ordinary kriging was used to estimate the blocks.

The Gara block size optimisation was run in domain 3000 where there are two different data density areas as a result of the grade control and exploration areas. Table14-28 details the QKNA results for Gara.

18^th September 2018 Page 154

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-28 QKNA Parameters from Gara

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
OP Grade Control	5	20	10	1	32.5	15	10	12	34	2	5	4
				2	65	29	13	10	26
				3	130	58	26	8	18
UG Grade Control	5	20	10	1	65	29	13	16	36	2	5	4
				2	130	58	26	14	32
				3	260	126	52	12	28
Exploration	7.5	30	15	1	120	30	10	16	34	2	5	4
				2	240	60	20	14	30
				3	580	120	40	12	26

Loulo 3

The block size was optimised using QKNA in both exploration and grade controlsub-domains as shown in Table14-29.

Table14-29 QKNA Parameters from Loulo 3

Domain	Block Size  (m)			Run	Search Radius  (m)			No. Samples		Discretisation			High- Grade Constraint  (g/t)	Constraint Search Range (m)
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
MZ2 GC	10	5	5	1	15	10	5	6	15	5	5	5
				2	30	20	10	8	15
				3	50	30	15	3	20
MZ2 Res	20	10	5	1	50	30	15	12	20	5	5	5
				2	100	60	30	10	24
				3	500	300	150	8	24				12	50 by 50 by 10
MZ1 GC	10	5	5	1	15	10	5	8	15	5	5	5
				2	30	20	10	6	15
				3	50	30	15	3	20
MZ1 Res	20	10	5	1	50	30	15	12	20	5	5	5
				2	100	60	30	10	24
				3	500	300	150	8	24				12	30 by 20 by 10
FW GC	10	5	5	1	15	10	10	12	20	5	5	5
				2	40	20	10	10	22
				3	80	40	20	8	24

Baboto

The search ellipsoid of Baboto is configured to have the same attitude as the variogram. However, in some areas the dip and bearing change to follow the mineralisation. The bearing and dip are setup in the block model as variables, when the plunge is according to the variogram. Table14-30 details the QKNA parameters for Baboto South GC domains.

Table14-30 QKNA Parameters from Baboto

Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
Baboto GC	3	6	2.5	1	20	18	10	8	16	2	3	2
				2	40	36	19	6	14
				3	80	72	38	4	12

18^th September 2018 Page 155

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gara West

Table14-31 present the QKNA from Gara West.

Table14-31 QKNA Parameters from Gara West

Domain	Block Size   (m)			Run	Search Radius (m)			No. Samples		Discretisation			High-Grade   Restriction   (g/t)	Restriction
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
10000	20	10	5	1	37	32	7	15	25	5	5	5
				2	74	64	14	15	30
				3	185	160	35	10	50
20000	20	10	5	1	70	35	5	15	25	5	5	5
				2	140	70	10	10	30
				3	350	175	25	10	50				8	70 by 35 by 5
30000	20	10	5	1	53	47	6	15	25	5	5	5
				2	106	94	12	15	30
				3	265	235	30	10	50				5	53 by 47 by 6
40000	20	10	5	1	81	32	6	15	25	5	5	5
				2	162	64	12	15	30
				3	405	160	30	10	50

Gounkoto

The QKNA is used to optimise the block size, number of samples, searches applied and discretisation. Nine orientation domains (MZ1 east dipping, MZ1 west dipping, MZ1 central west dipping, MZ2NW, MZ3, MZ4, P64W1, P64W2 and HW1) were selected for variography studies. A subset of results from QKNA are detailed in Table14-32.

18^th September 2018 Page 156

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-32 QKNA & Kriging Parameters for Gounkoto

Block Model Domain/Code	Estimation Domain Code	Drill Spacing Domain	Block Size (m)			Run	Search Radius (m)			No. Samples		Discretisation			High-Grade Constraint (g/t)	Constraint Search Range (m)			Firm Boundary Range Boundary D2/D3/D6			Comment
			X	Y	Z		Y	X	Z	Min	Max	X	Y	Z		Y	X	Z	Y	X	Z
MZ1 east dipping -  1001	1	GC	5	10	3.3	1	25	10	8	6	16	3	3	3	3.42	7	6	6	10	8	3	1/3 of variogram
						2	158	63	50	6	16	3	3	3					-	-	-
		AdvGC	7.5	15	4.95	1	50	40	16	8	18	3	3	3					20	16	6
						2	315	252	101	6	16	3	3	3					-	-	-
		Expl	15	30	9.9	1	100	80	31	12	24	3	3	3					40	32	12
						2	630	504	195	10	22	3	3	3					-	-	-
						3	945	756	293	8	20	3	3	3					-	-	-
	2	GC	5	10	3.3	1	25	10	8	6	16	3	3	3	96.43	7	6	6	10	8	3	1/3 of variogram
						2	158	63	50	6	16	3	3	3					-	-	-
		AdvGC	7.5	15	4.95	1	50	40	16	8	18	3	3	3					20	16	6
						2	315	252	101	6	16	3	3	3					-	-	-
		Expl	15	30	9.9	1	100	80	31	12	24	3	3	3					40	32	12
						2	630	504	195	10	22	3	3	3					-	-	-
						3	945	756	293	8	20	3	3	3					-	-	-
	4	GC	5	10	3.3	1	25	10	8	6	16	3	3	3	70.00	7	6	6	10	8	3	1/3 of variogram
						2	158	63	50	6	16	3	3	3					-	-	-
		AdvGC	7.5	15	4.95	1	50	40	16	8	18	3	3	3					20	16	6
						2	315	252	101	6	16	3	3	3					-	-	-
		Expl	15	30	9.9	1	100	80	31	12	24	3	3	3					40	32	12
						2	630	504	195	10	22	3	3	3					-	-	-
						3	945	756	293	8	20	3	3	3					-	-	-
	5	GC	5	10	3.3	1	25	10	8	6	16	3	3	3	37.35	7	6	6	10	8	3	1/3 of variogram
						2	158	63	50	6	16	3	3	3					-	-	-
		AdvGC	7.5	15	4.95	1	50	40	16	8	18	3	3	3					20	16	6
						2	315	252	101	6	16	3	3	3					-	-	-
		Expl	15	30	9.9	1	100	80	31	12	24	3	3	3					40	32	12
						2	630	504	195	10	22	3	3	3					-	-	-
						3	945	756	293	8	20	3	3	3					-	-	-
MZ1 central west dipping - 1001	6	GC	5	10	3.3	1	20	17	5	8	16	3	3	3	129.00	6	5	5	8	7	2	1/3 of variogram
						2	84	71	21	6	14	3	3	3					-	-	-
		AdvGC	7.5	15	4.95	1	40	33	10	8	16	3	3	3					16	13	4
						2	168	139	42	6	14	3	3	3					-	-	-
		Expl	15	30	9.9	1	80	67	22	12	24	3	3	3					32	27	8
						2	336	281	84	10	22	3	3	3					-	-	-
	3	GC	5	10	3.3	1	20	13	10	6	14	3	3	3	70.00	7	4	2	8	5	4	1/3 of variogram
						2	126	82	63	4	12	3	3	3					16	10	8
						3	504	328	252	2	10	3	3	3					32	20	16
						4	1008	655	504	2	8	3	3	3					64	40	32
		AdvGC	7.5	15	4.95	1	40	26	20	8	18	3	3	3					16	10	8
						2	252	164	126	6	16	3	3	3					32	10	16
		Expl	15	30	9.9	1	80	53	39	8	18	3	3	3					-	-	-
						2	504	334	246	6	16	3	3	3					-	-	-

18^th September 2018 Page 157

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Faraba

The QKNA has been completed on each variogram domain with the first pass of estimation, this analysis has been conducted to select an optimal estimation parameters with an aim to obtain an estimate grade comparable to the raw grade in term of the distribution and the similarity of the value (Table14-33).

Table14-33 Summary of QKNA for Faraba Domains

Domain	Block Size (m)			Run	Search   Radius (m)			No.   Samples		Discretisation			High- Grade   Constraint   (g/t)	Constraint   Search Range   (m)
	X	Y	Z		Y	X	Z	Min	Max	X	Y	Z
Main Zone (1000)	20	20	10	1	80	55	15	24	44	5	5	5
				2	160	110	30	22	44				2.7	80 by 55 by 15
				3	320	220	60	20	44				2.7	80 by 55 by 15
Hanging Wall (2000)	20	20	10	1	80	55	15	24	44	5	5	5
				2	160	110	30	22	44				2.0	80 by 55 by 15
				3	320	220	60	20	44				2.0	80 by 55 by 15
Footwall (3000)	20	20	10	1	80	55	15	24	44	5	5	5
				2	160	110	30	22	44				2.0	80 by 55 by 15

14.11 Block Models

Setup

Table14-34 details the typical block model variables and attributes that were coded to each block model either prior to, or during each interpolation run for OK models. These variables are typical of all the block models created at the Project.

Table14-34 Yalea Block Model Variables

Variables	Data Type	Default   Value	Description
domain	Integer (Integer * 4)	2000	9001=ysth, 9003=ynth, 9006= pp, 9005=p125, 9002=ysth2, 9004=ynth4, 9007=ysth3, 2000=waste, 1000=dyk
au_ok	Double (Real * 8)	0	estimated grade from kriging
density	Double (Real * 8)	0	density
sr	Float (Real * 4)	0	slope of regression
ns	Integer (Integer * 4)	0	number of sample used in estimation
kv	Float (Real * 4)	0	kriging variance
bv	Float (Real * 4)	0	Block variance = average sample variance - block sample variance
ke	Float (Real * 4)	0	Kriging efficiency
lm	Float (Real * 4)	0	Lagrange multiplier
dist_ans	Float (Real * 4)	0	Anisotropic distance to nearest sample
dist_cat	Float (Real * 4)	0	Cartesian distance to closest Sample
no_hole	Integer (Integer * 4)	0	number of holes in estimation
no_pass	Integer (Integer * 4)	0	no of estimation run
wt_sum	Float (Real * 4)	0	Sum of weights
wt_mean	Float (Real * 4)	0	Weight of the mean
oxidation	Integer (Integer * 4)	0	oxidation 0=air 30= fresh 20= transition 10= oxide
rescat	Integer (Integer * 4)	4	resource classification 0=AIR, 1=Measured, 2=Indicated 3=Inferred 4 = unclass
depletion	Integer (Integer * 4)	0	depletion on model 100 = Open Pit depleted 200 = UG
lith	Integer (Integer * 4)	2000	9000=ore, 2000=fw waste, 2500=hw waste, 1000=dyk, 0=AIR
redox	Integer (Integer * 4)	0	redox front 1= oxidised 2 = reduce
au_orig	Float (Real * 4)	0	original estimated grade before depletion
mined_date	Integer (Integer * 4)	0	mined zone YYMMDD = Year and Month e.g. (121001)
gd_ans	Float (Real * 4)	0	grade of the closest sample (anisotropic)
gd_cat	Float (Real * 4)	0	grade of the closest sample (cartesian)

18^th September 2018 Page 158

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Variables	Data Type	Default   Value	Description
nw_sum	Float (Real * 4)	0	sum of negative weights
pw_sum	Float (Real * 4)	0	sum of positive weights
kw_min	Float (Real * 4)	0	Minimum Kriging weight
kw_max	Float (Real * 4)	0	Maximum Kriging Weight
avd_ans	Float (Real * 4)	0	average anisotropic distance to sample derived from search ellipsoid
wvd_ans	Float (Real * 4)	0	weighted average anisotropic distance to sample derived from search ellipsoid
avd_cat	Float (Real * 4)	0	cartesian average distance to samples
wvd_cat	Float (Real * 4)	0	weighted cartesian average distance to samples
flag_ore	Integer (Integer * 4)	0	Flag = 1 when block contains ore
cu_ok	Double (Real * 8)	0	estimated cu grade from kriging
sr_cu	Double (Real * 8)	0	slope of regression cu
ns_cu	Integer (Integer * 4)	0	number of cu sample used in estimation
kv_cu	Double (Real * 8)	0	kriging variance
ke_cu	Float (Real * 4)	0	Kriging efficiency
dist_ans_cu	Float (Real * 4)	0	Anisotropic distance to nearest sample
dist_cat_cu	Float (Real * 4)	0	Cartesian distance to closest Sample
no_hole_cu	Integer (Integer * 4)	0	number of holes in cu estimation
no_pass_cu	Integer (Integer * 4)	0	no of estimation run
gd_ans_cu	Float (Real * 4)	0	cu grade of the closest sample (anisotropic)
gd_cat_cu	Float (Real * 4)	0	cu grade of the closest sample (cartesian)
cu_orig_cu	Float (Real * 4)	0	original estimated cu grade before depletion
nw_sum_cu	Float (Real * 4)	0	sum of negative weights
avd_ans_cu	Float (Real * 4)	0	average anisotropic distance to sample derived from search ellipsoid
wvd_ans_cu	Float (Real * 4)	0	weighted average anisotropic distance to sample derived from search ellipsoid
avd_cat_cu	Float (Real * 4)	0	cartesian average distance to samples
wvd_cat_cu	Float (Real * 4)	0	weighted cartesian average distance to samples
flag_ore_cu	Float (Real * 4)	0	Flag = 1 when block contains cu
ars_ok	Double (Real * 8)	0	estimated as grade from kriging
sr_ars	Double (Real * 8)	0	slope of regression ars
ns_ars	Integer (Integer * 4)	0	number of ars sample used in estimation
kv_ars	Double (Real * 8)	0	kriging variance
ke_ars	Float (Real * 4)	0	Kriging efficiency
dist_ans_ars	Float (Real * 4)	0	Anisotropic distance to nearest sample
dist_cat_ars	Float (Real * 4)	0	Cartesian distance to closest Sample
no_hole_ars	Integer (Integer * 4)	0	number of holes in ars estimation
no_pass_ars	Integer (Integer * 4)	0	no of estimation run
gd_ans_ars	Float (Real * 4)	0	ars grade of the closest sample (anisotropic)
gd_cat_ars	Float (Real * 4)	0	ars grade of the closest sample (cartesian)
cu_orig_ars	Float (Real * 4)	0	original estimated cu grade before depletion
wvd_cat_ars	Float (Real * 4)	0	weighted cartesian average distance to samples
flag_ore_ars	Float (Real * 4)	0	Flag = 1 when block contains ars
nw_sum_ars	Float (Real * 4)	0	sum of negative weights
wvd_ans_ars	Float (Real * 4)	0	weighted average anisotropic distance to sample derived from search ellipsoid
avd_cat_ars	Float (Real * 4)	0	cartesian average distance to samples
avd_ans_rec	Float (Real * 4)	0	recovery cartesian anisotropic distance to samples
wvd_cat_rec	Float (Real * 4)	0	weighted cartesian average distance to samples
avd_cat_rec	Float (Real * 4)	0	recovery cartesian average distance to samples
flag_rec	Float (Real * 4)	0	recovery flag
dist_ans_rec	Double (Real * 8)	0	recovery anisotropic distance to nearest sample
dist_cat_rec	Double (Real * 8)	0	recovery Cartesian distance to closest Sample
no_pass_rec	Float (Real * 4)	1	recovery number off estimation run
gd_ans_rec	Double (Real * 8)	0	recovery of the closest sample (anisotropic)
gd_cat_rec	Double (Real * 8)	0	recovery of the closest sample (cartesian)
no_hole_rec	Double (Real * 8)	0	recovery number of hole
ns_rec	Double (Real * 8)	0	recovery number of sample
recovery	Double (Real * 8)	0	process recovery %
contained_oz	Double (Real * 8)	0	Block contained ounces = block volume * density * gold grade
recoverable_oz	Double (Real * 8)	0	Block recoverable ounces = block contained ounces * process recovery

18^th September 2018 Page 159

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo

Yalea

Different block sizes are used based on the data density and from that different estimation subdomains were derived.

The GC sub domains refers to the area where the drilling spacing is between 10 m to 30 m i.e. grade control drilled areas. These are labelled using a value of in 100 in the domain code e.g. 9102. The Advanced GC sub domains refer to the area where the drilling spacing is between 30 m to 80 m spacing. This comprises the wider spaced areas which are mainly exploration drilled areas with some underground grade control area. These are labelled with a value of a 200 in the domain code. e.g. 9202.

The exploration sub domains are exploration areas with drill spacing greater than 80 m which is either classified as Inferred or unclassified. These sub domains are coded with a 300 in the domain code e.g. 9302.The broad drill hole density classifications are then further split into the estimation domains based upon a combination of geological, orientation and drill spacing sub domains (Figure14-26).

The Yalea block model name is “20171020_yalea_uggc_model.bmf” and is a sub cell model generated from Vulcan. The block model extents variables are presented in Table14-35. Table14-36 outlines the Yalea search orientations andsub-domain nomenclature.

Figure14-26 Yalea 2017 Drill hole Density Sub Domains

18^th September 2018 Page 160

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-35 Yalea Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation(Z)
Minimum Coordinates		240160	1440320	-1000
Maximum Coordinates		241110	1443590	240
Parent Block Size (m)		10	30	20
Sub Cell Size (m)		0.5	1.5	1
GC Domain Block Size (m)		5	15	10
Adv GC Domain Block Size (m)		8	24	16
Rotation	Bearing (°)	90
	Plunge (°)	0
	Dip (°)	0

Table14-36 Yalea Estimation Domain Orientation

Geological Domain	Estimation Domains	Domain	Domain Orientation
			Bearing (°)	Plunge (°)	Dip (°)
9001 (Yalea South)	9101	GC	357	9	117
	9201	AdvGC	357	9	117
	9301	Exp	357	9	117
9002 (Yalea South)	9102	GC	0	9	-80
	9202	AdvGC	5	9	-90
	9302	Exp	5	9	-90
9007 (Yalea South) unclassified	9317		0	9	110
	9327		0	9	90
	9337		10	9	60
	9347		10	9	90
9003 (Yalea North)	9103	GC	4.577	-8.901	117.356
9004 (Yalea North)	9104	GC	0	-8.901	-85
	9204	AdvGC	0	-8.901	-85
9005 (P125)	9105	GC	3	0	-75
9006 (‘Purple Patch’)	9106	GC	183	-15	90
	9156	GC	183	-15	90
	9206	AdvGC	183	-15	90

Gara

At Gara the same approach is used by sub domaining and domain nomenclature to account for the variation in drill spacing from 10 m to 120 m. The grade control area is sub split into the open pit grade control where the drill spacing is approximately 10 m and the underground drill spacing area where the drill spacing is less than or equal to 30 m. The area where the drilling is between 30 m to 90 m is defined as the resource exploration area.

These broad domains are then further split into the estimation domains based upon a combination of geological, orientation and drill spacing sub domains. These estimation domains are used to define the estimation parameters are displayed in Figure14-27. The Gara block model name is “20171120_gara_uggc_model_EXT_SOUTH.bmf” and is a sub cell model generated from Vulcan. Table14-37 outlines the Gara block model extents. Table14-38 outlines the Yalea search orientations andsub-domain nomenclature.

18^th September 2018 Page 161

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure14-27 Gara 2017 Estimation Domains

18^th September 2018 Page 162

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-37 Gara Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation(Z)
Minimum Coordinates		237400	1445500	-850
Maximum Coordinates		238502.5	1448710	185
Parent Block Size (m)		7.5	30	15
Sub Cell Size (m)		0.5	2	1
GC Domain Block Size (m)		5	20	10
Rotation	Bearing (°)	90
	Plunge (°)	0
	Dip (°)	0

Table14-38 Gara Estimation Domain Orientation

Geological Domain	Estimation   Domains	Comment	Domain Orientation
			Bearing (°)	Plunge (°)	Dip (°)
3000	3101	Open pit GC	216.102	-25.659	-56.31
	3102	UG GC	212.102	-25.659	-56.31
	3200	Exploration	210.102	-25.659	-76.31
4000	4100	UG GC	195	-20.705	139.107
	4200	Exploration	210	-20.705	134
	4205	Exploration South extension	202	-20.705	124
4605
5000	5101	Open pit GC	295	0	-150
	5102	UG GC	300	0	-150
6000	6101	Open pit GC	38.539	13.566	115.769
	6102	UG GC	26.539	13.566	125.769
7000	7101	Open pit GC	30.703	18.829	105.055
	7102	UG GC	26.703	18.829	105.055
	7200	OP Exploration	10	0	-60
7400 & 7600	7300	OP Exploration	8	0	-40
	7400		10.703	0	7
	7500		8	0	-60
	7600		15	10	35
	7700		7	1	10
	7800		2	1	90
7605	4205	Exploration South extension	202	-20.705	124
	7405	OP Exploration	10.703	10	7
	7505		5	10	90
	7605		15	18.829	35
	7705		7	12	10
	7805		-5	1	60
8000	8101	UG GC	20	18	110
	8102	UG GC	355	18	80
	8103	UG GC	20	18	110
8000 & 8400	8201	Exploration	10	18	115
8000 & 8900	8202	Exploration	355	18	160
8000	8203	Exploration	20	18	115
9000	9100	UG GC	5	18.783	100.3
	9200	Exploration	5	18.783	100.3
9400 & 9600
9605	9205	Exploration South extension	12	18.783	90.3
10000	10100	UG GC	5	18.783	100.3

18^th September 2018 Page 163

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo 3

All Loulo 3 block sizes are optimised to be approximately half of the relevant along strike drill spacing in each domain. Subsequently, block sizes have smaller X axis relative to Y axis due to the elongate thin nature of the mineralisation in areas.

The Main Zone and South Zone are bothsub-domained into two sub domains: exploration or grade control based on drill spacing. This approach enables application of different search ranges and use of different block sizes within each sub domain (Table14-39). Allsub-domains use soft boundaries as there is no geological reason to constrain the resource estimate from using sample support from the other domains. Table14-40 details the search ellipsoid orientation and high-grade constraining applied.

The block model name for Loulo 3 is “150101_LOULO 3_RESCAT.bmf” which is a sub cell model generated from Vulcan.

Table14-39 Loulo 3 Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation(Z)
Minimum Coordinates		240975	1445200	-450
Maximum Coordinates		242075	1447700	200
Parent Block Size (m)		10	20	5
Sub Cell Size (m)		2.5	2.5	0.625
GC Domain Block Size (m)		5	10	5
Rotation	Bearing (°)		100
	Plunge (°)		0
	Dip (°)		0

Table14-40 Loulo 3 Search Ellipse Orientations and HG Constraining

Domain	Sub-Domain	Pass Number	Search Ellipsoid
			Azi (°)	Plunge (°)	Dip (°)
MZ2	GC	1	20	0	-50
		2	20	0	-50
		3	20	0	-50
	RES	1	20	0	-50
		2	20	0	-50
		3	20	0	-50
MZ1	GC	1	350	0	-50
		2	350	0	-50
		3	350	0	-50
	RES	1	350	0	-50
		2	350	0	-50
		3	350	0	-50
			350	0	-50
			350	0	-50
FW	GC	1	350	0	-50
		2	350	0	-50
		3	350	0	-50

Baboto

The Baboto 2017 block sizes have small X axis relative to Y axis due to the elongated thin nature of the mineralisation in areas. The block model name for Baboto resource model is “BB_GC_170514_v2.bmf” which is a sub cell model generated from Vulcan. The extents of the block model Table14-41.

18^th September 2018 Page 164

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Baboto South and Centre aresub-domained into two sub domains of exploration and grade control based on drill spacing (Figure14-28). This approach enables application of different search ranges and use of different block sizes within each sub domain (Table14-42). All subdomains will use soft boundaries as there is no geological reason to constrain the resource estimate from using sample support from other domains.

Figure14-28 Baboto Estimation Domains Looking WNW

Table14-41 Baboto South and Centre Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation(Z)
Minimum Coordinates		244030	1452985	-200
Maximum Coordinates		245510	1457285	310
Parent Block Size (m)		5	15	10
Sub Cell Size (m)		0.25	0.5	0.25
GC Domain Block Size (m)		3	6	2.5
Rotation	Bearing (°)		90
	Plunge (°)		0
	Dip (°)		0

18^th September 2018 Page 165

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-42 Baboto South and Centre Estimation Domains

Model	Lith in BM	Domain in BM	Estimation Domain	Search Ellipsoid
				Azimuth (°)	Plunge (°)	Dip (°)
Baboto South Main	1001	GC	1111	344	0	84
			1121	357	0	82
			1131	340	0	81
			1141	356	0	83
		Expl	1211	344	0	84
			1221	357	0	82
			1231	340	0	81
			1241	356	0	83
Baboto South East	1002	GC	11021	344	0	89
			11022	346	0	88
			11023	360	0	77
			11024	359	0	76
			11025	356	0	79
			11026	359	0	84
			11027	347	0	74
			11028	349	0	89
		Expl	12021	344	0	84
			12022	357	0	82
Baboto South HW	1003	GC	1103	340	0	81
		Expl	1203	356	0	83
Baboto South FW	1004	GC	1104	347	0	74
		Expl	1204	347	0	74
Baboto Central	2001	GC/Expl	2001	3	-9	-100

Gara West

The Gara West block sizes are optimised to be approximately half of the relevant along strike drill spacing. The Gara West is mostly drilled on regular spacings across all domains, therefore all domains have the same block size. The extents of the Gara West block model are detailed in Table14-43.

There are four domains modelled at Gara West, however, there is significant small to medium variations in the strike and dip in each individual domain and as such, each domain has beensub-domained (Figure14-29). This approach enables application of different search orientations within each sub domain (Table14-45). Allsub-domains will use soft boundaries as there is no geological reason to constrain the resource estimate from using sample support from other domains.

Table14-43 Gara West Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation (Z)
Minimum Coordinates		237400	1446100	-850
Maximum Coordinates		238510	1448710	185
Parent Block Size (m)		10	20	5
Sub Cell Size (m)		1	2	0.5
Rotation	Bearing (°)		90
	Plunge (°)		0
	Dip (°)		0

18^th September 2018 Page 166

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-44 Gara West Sub Domain Ellipse Orientations

Domain	Estimation Domain	Search Ellipsoid
		Azimuth (°)	Plunge (°)	Dip (°)
10000	1	24	0	67
	2	26	0	60
20000	1	14	0	64
	2	27	0	68
	3	26	0	68
	4	13	0	58
	5	15	0	59
	6	15	0	53
30000	1	16	0	65
	2	22	0	71
	3	18	0	59
40000	1	44	0	69
	2	24	0	69
	3	20	0	57

Figure14-29 Gara West EstimationSub-Domains

The block model name for Gara West is ‘150101_GARAWEST_RESCAT.bmf’ which is a sub cell model generated from Vulcan.

Gounkoto

The plunge in the primary east dipping domains is much shallower (9°) than the plunge within west and central west dipping mineralisation as a function of the plunge of FW fingers within the west dipping domains. The changing strike and dip of the Gounkoto Main Zone structure required that estimation domains weresub-domained to separate the east and west dipping mineralisation foliation. Estimation parameters were optimised on the parent domains and the resulting optimisations were applied to the sub domains. The search ellipse was orientated individually for each sub domain where necessary. The Gounkoto block model extents are detailed in Table14-45.

18^th September 2018 Page 167

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-45 Gounkoto Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation(Z)
Minimum Coordinates		239595	1422840	-675
Maximum Coordinates		241545	1425930	176.4
Parent Block Size (m)		15	30	9.9
Sub Cell Size (m)		1.25	2.5	0.825
Advance GC Block Size (m)		7.5	15	4.95
GC Block Size (m)		5	10	3.3
Rotation	Bearing (°)		90
	Plunge (°)		0
	Dip (°)		0

A summary of the Gounkoto search ellipse parameters and the high-grade constraints applied to some domains is detailed in Table14-46.

Table14-46 Summary of the Gounkoto Orientations and HG Constraint

Block Model Domain Code	Estimation Domain Code	Search Ellipse Orientation
		Strike (°)	Dip (°)	Plunge (°)
MZ1 east dipping - 1001	1	360	110	-9
	2	345	125	-9
	9	360	105	0
	4	360	110	-9
	5	345	120	-9
MZ1 central west dipping - 1001	3	10	80	0
	6	345	120	0
MZ2NW - 1002	OP	350	70	0
	UG	350	70	0
MZ2NW - 10022	-	30	120	-9
10023	-	360	100	-9
MZ3 - 1003	7 OP	350	110	-8
	7 UG	350	110	-8
	8 OP	20	120	-8
	8 UG	20	120	-8
	13	350	110	-8
ID - 10032	-	10	110	0
MZ4 - 1004	-	350	80	0
10042	-	360	100	0
P64W1 (P64WNS) - 1005	-	4	90	0
P64W2 (P64WNE) - 1006	-	40	80	0
P64W3 (P64WNS) - 1007	-	6	80	0
P64W1 (P64WNS) - 1008	P64W HW1
P64W1 (P64WNS) - 1009	P64W HW2
P64EW1 (P64WNS) - 10051	10	4	90	0
	11	310	90	0
P64EW1 (P64WNS) - 10052	12	310	90	0
HW1 - 2001	-	5	120	0
HW1 - 2002	-	5	120	0
HW1 - 2003	-	5	120	0
HW1 - 2004	-	5	120	0
MZ1 West Dipping - 3001 3002 3003 3004 3005 3006	-	360	90	0
3007 3008 3009 3010 3011
MZ1 West Dipping - 4001 4002 4003 4004 4005	-	15	80	0
Waste (9999)	-	360	120	0

The east and west dipping domain boundary has been modelled using a surface which is interpreted to follow the dip change in MZ1 with all the data available (from pit mapping and from diamond hole). The domain that is situated between these structural domains has been created

18^th September 2018 Page 168

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

using evidence from the drill holes where the original fabrics are dipping away from each other in a highly altered zone.

Faraba

The three domains at Faraba (Main Zone 1000, Hanging Wall 2000, and Footwall 3000) were not sub domained. The search ellipsoid was orientated individually for each domain. The block model extents for Faraba are presented in Table14-47.

Table14-47 Faraba Block Model Extents

Block Extents		Easting(X)	Northing(Y)	Elevation(Z)
Minimum Coordinates		241930	1421290	-350
Maximum Coordinates		242730	142550	-250
Parent Block Size (m)		20	20	10
Rotation	Bearing (°)		90
	Plunge (°)		0
	Dip (°)		0

Table14-48 summarises the azimuth, dip and plunge along with the high-grade constrains applied to each domain during ordinary kriging.

Table14-48 Modelled Semi-Variogram Parameters for Faraba

Domain	Estimation Domain	Search Ellipsoid
		Azi (°)	Dip (°)	Plunge (°)
Main Zone (1000)	1	358	-57	0
	2	349	-55	0
	3a	301	-64	0
	3b	301	-83	0
	4a	359	-64	0
	4b	359	-83	0
	5a	304	-64	0
	5b	304	-83	0
	6a	4	-64	0
	6b	4	-83	0
Hanging Wall (2000)	1a	359	-55	0
	1b	359	-71	0
	2	351	-53	0
Footwall (3000)	1	356	88	0
	2	326	-86	0
	3	345	-88	0
	4	293	-88	0
	5	360	-77	0
	6	360	81	0

14.12 Resource Classification

Under the CIM definitions (CIM (2014) Standards on Mineral Resources and Reserves Definitions and Guidelines, August 2000), and the JORC (2012) Code, Measured Mineral Resources require that “quantity, grade, density, shape, and physical characteristics need to be established with confidence sufficient to allow the appropriate application of technical and economic parameters” such that production planning and the evaluation of the economic viability of the deposit is possible.

18^th September 2018 Page 169

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

In the case of Measured and Indicated Resources, the level of confidence should be sufficient to allow for the application of appropriate technical and economic parameters, mine planning, and economic evaluation.

Resource Classification applied to Loulo-Gounkoto Mineral Resources has taken this into consideration and the classifications are based on a combination of geological continuity, data density, variogram range continuity as well as estimation quality in form of slope of regression (SR) and kriging efficiency (KE). This was carried out by displaying the estimated blocks (SR and KE) together with the supporting data as guide.

To increase the continuity of the Mineral Resource, classification shapes are built using the criteria that reflects the confidence in the estimation for Measured, Indicated, and Inferred classification. These shapes are used flag the block model. Blocks that do not meet these criteria are flagged as unclassified and used for exploration and resources conversion evaluation purpose.

Loulo

The criteria for determining the classification of each deposit is outlined in Table14-49.

Table14-49 Loulo Classification Criteria

Class	Criteria	Yalea	Gara	Loulo 3	Baboto	Gara West
Measured	Maximum Drill Spacing (m)	30	30	5 by 10	5 by 10	N/A
	Geological Continuity	6 Sections	6 Sections	6 Sections	6 Sections
	Min Samples	8	8	8	8
	Minimum Slop of Regression	0.8	0.8	0.8	0.8
Indicated	Maximum Drill Spacing (m)	30 to 80	40 to 80	20 to 80	20 to 60	10 to 40
	Sections of Geological Continuity	Good	Good	Good	Good	Good
	Min Samples	6	6	6	6	6
	Minimum Slop of Regression	0.6	0.6	0.6	0.6	0.6
Inferred	Maximum Drill Spacing (m)	Over 80	Over 80	Over 80	Over 80	Over 40
	Sections of Geological Continuity	-	-	-	-	-
	Min Samples	4	4	4	4	4
	Minimum Slop of Regression	-	-	-	-	-

18^th September 2018 Page 170

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Yalea

Figure14-30 illustrates the Yalea Mineral Resources classification.

Figure14-30 Yalea Mineral Resources Classification with Estimation Composites

Gara

Figure14-31 shows the Gara Mineral Resources classification.

Figure14-31 Gara Mineral Resources Classification with Estimation Composites

Loulo 3

Figure14-32 shows the Loulo 3 Mineral Resources classification.

18^th September 2018 Page 171

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure14-32 Loulo 3 Mineral Resource Classification with Estimation Composites

Baboto

Figure14-33 displays the Baboto South and Central Mineral Resources classification.

Figure14-33 Baboto 2017 Mineral Resource Classification Surfaces with Estimation Composites

Gara West

Figure14-34 shows the Gara West Mineral Resources classification.

Figure14-34 Gara West 2017 Mineral Resource Classification Surfaces with Estimation Composites

18^th September 2018 Page 172

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

The criteria for determining the classification of each deposit is outlined in Table14-50. Slope of regression plots, the variogram range and geological continuity were also used.

Table14-50 Gounkoto Classification Criteria

Class	Criteria	Gounkoto Open Pit	Gounkoto   Underground	Faraba
Measured	Maximum Drill Spacing (m)	12.5 by 12.5   (6 b 6 Finger Zone)	N/A	12.5 by 12.5
	Geological Continuity
	Minimum Slope of Regression	0.7
Indicated	Maximum Drill Spacing (m)	30	40 by 30	30
	Sections of Geological Continuity
	Minimum Slope of Regression	0.5	0.5	0.5
Inferred	Maximum Drill Spacing (m)	<100	<100	<100
	Sections of Geological Continuity

Figure14-35 illustrates an example of the detailed classification on MZ1.

Figure14-35 Gounkoto July 2017 MZ1 Classification

18^th September 2018 Page 173

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Faraba

Figure14-36 illustrates the Mineral Resource classification at Faraba.

Figure14-36 Faraba Classification Within Pit

14.13 Block Model Depletion

Every block model was depleted against the 5 m Regional LIDAR DTM surface that was completed by Randgold in 2010. Where surface mining has occurred, updated LIDAR surveys have been completed on an annual basis.

Underground stope scans are used to report depletion from the block models which is then applied to the Mineral Resource reporting.

14.14 Block Model Validation

Once the block model has been classified, checks were undertaken to ensure the block models and estimated grades which can be used to indicate any major errors during the estimation process as well as testing the precision, accuracy, and any bias of the estimated grade.

The block model was validated using the following steps.

Volume Reconciliation

A volume reconciliation between the block model estimation domains and related wireframes is undertaken. Table14-51 summarises the variances between the wireframe and block model volumes across all Yalea and Gara.

18^th September 2018 Page 174

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-52 summarised the volume reconciliation for Gounkoto. The overall variance is shown to be excellent (<1%).

Table14-51 Gara and Yalea Volume Reconciliation

Model	Zone	BM Volume  (m3)	WF Volume  (m3)	Variance (m3)	Variance
Yalea	9001	2,281,297	2,280,033	1,264	-0.10%
	9002	2,353,633	2,353,566	67	0.00%
	9003	2,974,154	2,975,444	-1,290	0.00%
	9004	2,984,667	2,985,095	-428	0.00%
	9005	593,775	590,372	3,403	-0.60%
	9006	1,948,751	1,948,528	223	0.00%
	Total	13,136,278	13,133,038	3,240	0.00%
Gara	UG_3000	3,029,424	3,029,454	-30	0.00%
	UG_4000	942,758	943,007	-249	-0.03%
	UG_5000	179,827	179,780	47	0.03%
	UG_6000	96,365	96,430	-65	-0.07%
	UG_7000	367,738	367,499	239	0.06%
	UG_8000	3,460,869	3464507	-3,638	-0.11%
	UG_9000	1,410,918	1,410,905	13	0.00%
	UG_9600	1,785,589	1,785,719	-130	-0.01%
	UG_4600	329,480	329,448	32	0.01%
	UG_4605	456,657	456,607	50	0.01%
	UG_7400	661,336	661,768	-432	-0.07%
	UG_7600	850,408	850,939	-531	-0.06%
	UG_7605	520,549	521,205	-656	-0.13%
	UG_8400	680,790	680,678	112	0.02%
	UG_8900	955,465	955,506	-41	0.00%
	UG_9400	246,830	246,899	-69	-0.03%
	UG_9605	1,528,520	1,528,884	-364	-0.02%
	Total	17,503,523	17,509,236	-5,713	-0.03%

18^th September 2018 Page 175

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-52 Gounkoto Volume Reconciliation

Zone	WF Volume (m3)	BM Volume (m3)	Variance (m3)	Variance
1001	9,832,409	9,830,777	-1,631	-0.02%
1002	2,148,367	2,151,708	3,341	0.16%
1003	5,067,423	5,068,942	1,519	0.03%
1004	366,963	367,388	425	0.12%
1005	520,817	520,833	15	0.00%
1006	453,990	453,920	-70	-0.02%
1007	86,280	86,292	12	0.01%
1008	18,629	18,459	-169	-0.91%
1009	9,400	9,444	44	0.47%
2001	6,018,477	6,015,921	-2,556	-0.04%
2002	19,634	19,645	11	0.06%
2003	241,036	241,142	106	0.04%
2004	21,434	21,816	382	1.78%
3001	370,057	370,497	440	0.12%
3002	78,526	78,501	-25	-0.03%
3003	962,899	960,831	-2,068	-0.21%
3004	278,669	278,363	-306	-0.11%
3005	73,960	73,474	-487	-0.66%
3006	14,161	14,064	-97	-0.68%
3007	51,591	51,044	-547	-1.06%
3008	2,020	1,892	-128	-6.34%
3009	75,902	76,859	957	1.26%
3010	56,260	55,876	-384	-0.68%
3011	22,933	22,512	-421	-1.84%
4001	31,425	31,443	18	0.06%
4002	17,674	17,699	25	0.14%
4003	49,242	49,273	32	0.06%
4004	54,831	54,757	-74	-0.13%
4005	175,925	175,944	19	0.01%
10022	124,454	124,895	441	0.35%
10023	3,865	3,836	-29	-0.75%
10032	9,505,192	9,502,481	-2,710	-0.03%
10042	13,474	13,961	486	3.61%
10051	849,080	850,516	1,435	0.17%
10052	16,413	16,258	-155	-0.94%
Total	37,633,413	37,631,264	-2,149	-0.01%

Negative Kriging Weight Check

A check of the number of the blocks estimated using the negative kriging weight is completed. If blocks are observed to have a material impact as a result of negative kriging weights, the block grade is reset to that of the anisotropic nearest block grade. At Gounkoto, 50 blocks in MZ2, MZ3, and P64E1 grade control area were estimated using negative kriging weight and therefore reset to the nearest anisotropic block grade.

Data Comparison

A comparison between the data minimum, maximum, mean and the estimation mean for each of the domains (within the open pit or underground reporting areas) is created. This is completed to check for possible over or under estimation. An example of the composite and block model grade comparison for Gounkoto Open Pit is presented in Table14-53.

18^th September 2018 Page 176

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-53 Gounkoto Open Pit Data Comparison

Zone	Min Au (g/t)		Max Au (g/t)		Mean Au (g/t)
	Comp	BM	Comp	BM	Comp	BM
MZ1	0.005	0.009	156.96	74.72	5.13	5.66
MZ2	0.005	0.027	73.25	48.86	7.65	5.74
MZ3	0.005	0.020	98.30	34.97	5.78	5.62
MZ3HW	0.031	0.112	1.56	0.69	0.30	0.35
MZ4	0.005	0.064	51.77	25.22	3.56	3.13
MZ4FW	0.290	0.648	1.39	2.11	0.79	0.88
HW (1 to 4)	0.005	0.007	63.80	26.31	1.75	1.98
FWFE (1 to 11)	0.005	0.042	65.22	19.11	2.71	2.71
FWNE (1 to 5)	0.005	0.439	6.76	2.11	1.46	1.24
P64W (1 to 3) +P64HW (1 to2)	0.005	0.076	98.21	21.21	1.93	1.98
P64E (1 to2)	0.005	0.023	14.27	8.48	1.45	1.53
Total	0.005	0.007	156.96	74.72	4.06	3.91

The Highest variance between the block and sample in MZ2, but after declustering the data the sample mean become 5.88 g/t Au which is close to the block mean grade. MZ3HW high variance is link to the wide space drilling that is why this zone is classified as Inferred, MZ4 high variance is driven by the 100 m extension in the south with wide spacing drilling, this extension is classified as Inferred Resources.

Swath Plots

Swath plots are created for each geological domain to validate the estimated grade variability compared to the composite along the Y axis which is along strike of the mineralisation The X and Z directions are also reviewed. This is to check that the model estimate follows the trends seen in the data and that there is no bias with over or under estimation. An example of the swath plots is presented in Figure14-37 which illustrates swath plot of the Y axis for 9106 of the ‘Purple Patch’ and Figure14-38 illustrates a swath plot of the Y axis for Gara 3000.

18^th September 2018 Page 177

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure14-37 Swath Plot of Yalea Grade Control Area of 9006

Figure14-38 Swath Plot Along Y Axis of 3101 (3000 GC Area)

18^th September 2018 Page 178

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Alternative Estimations

Block models are also estimated using nearest neighbour analysis as a check estimate against OK. Resource reports are generated, and visual checks are undertaken to compare them to the OK models.

Visual Checks

A comprehensive visual check is undertaken comparing the data to the block estimates to check for an acceptable correlation, including checking on local high-grade trends on the block estimated is supported by data as well as to compare previous block model iterations to the new block model. An example of the Yalea and Gara visual checks are illustrated in Figure14-39 and Figure14-40.

Figure14-39 Yalea 2017 Resource Model Looking East Regularised for Presentation Purposed and

Composite Samples Overlain onto Display Relative Accuracy of the Estimate

Figure14-40 Gara 2017 Resource Model Looking East Regularised for Presentation Purposed and

Composite Samples Overlain onto Display Relative Accuracy of the Estimate

18^th September 2018 Page 179

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.15 ResourceCut-Off Grades

Loulo

Thecut-off grade calculations for the Loulo open pit and underground operations are presented in Table14-54 and Table14-55.

Table14-54 2017 Loulo Open PitCut-Off Grade Calculations

Parameter	Unit	Symbol	Loulo 3	Baboto	Gara West
Gold Price	$/oz.	Gp	1500	1500	1500
Royalty	%	R	6%	6%	6%
Selling cost	%	S	0%	0%	0%
Net Gold Price	$/oz.	Ng	1410	1410	1410
Met Recovery	%	Rec	92%	93%	89%
Dilution	%	Dil	10%	10%	10%
Ore Loss	%	Loss	3%	3%	3%
Mining Cost - Contractor	$/t mined	MCC	3.59	3.18	3.64
Mining Cost - Owner’s	$/t mined	MCO	0.06	0.06	0.06
Mining Cost - GC	$/t mined	MCGC	0.07	0.07	0.2
Total Mining Cost	$/t mined	TMC	3.72	3.31	3.9
Strip Ratio	Waste/Ore	SR	20.56	4.24	7.44
G&A	$/t milled	G_A	8.8	7.8	8.8
Ore Crushing & hauling	$/t milled	rd	0	3.63	0
Mining	$/t milled	Cr	80.12	17.33	32.89
Process Plant	$/t milled	Cp	20.6	15.2	20.6
Maintenance/Engineering	$/t milled	Mp	0	0	0
Total Operating Costs	$		109.52	43.96	62.29
Full GradeCut-off	g/t		2.71	1.07	1.59
MarginalCut-off Grade	g/t		0.73		0.75

18^th September 2018 Page 180

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-55 2017 Loulo UndergroundCut-Off Grade Calculations

Parameter	Unit	Yalea	Gara
Production
Life of Mine Production 1	Mt	14.09	13.83
Capital Cost 2	$M	108.03	114.13
Revenue Adjustment
NSR Royalty (Royal Gold: NSR)3	of total oz produced	6.00%	6.00%
Refining and Selling Expense4	$/Ounce	$-	$-
Admin & Processing Costs
Metallurgical Recovery5	%	84.47%	94.20%
Direct Processing6	$/Ton	$17.80	$18.20
Tailings Disposal	$/Ton	$-	$-
Ore Haulage	$/Ton	$-	$-
Administration and Overhead7	$/Ton	$7.80	$7.70
Total Other Cost	$/Ton	$25.60	$25.90
Mining Costs
Dilution	%	11.50%	15.60%
Mining ore loss	%	2.00%	4.00%
OPEX Development8	$/Ton	$5.00	$3.29
OPEX Stoping9	$/Ton	$12.15	$11.79
Backfill10	$/Ton	$11.58	$10.51
Fixed Cost11	$/Ton	$13.98	$14.63
Grade control12	$/Ton	$3.38	$3.40
Total Mine Operating Cost	$/Ton	$46.09	$43.63
Mine Sustaining Capital	$/Ton	$7.67	$8.25
All in Cost	$/Ton	$53.76	$51.88
BreakevenIn-situCut-off Grade	g/t	2.36	2.19
Breakeven Mined Cut-off Grade	g/t	2.07	1.82
MarginalIn-situCut-off Grade	g/t	2.04	1.89
Marginal MinedCut-off Grade	g/t	1.74	1.55

Gounkoto

The Gounkoto and Farabacut-off grade calculations for open pit are presented in Table14-56.

Table14-56 Gounkoto Open Pit 2017Cut-Off Grade Calculations

Parameter	Unit	Symbol	Gounkoto	Faraba
Gold Price	$/oz.	Gp	1500	1,500
Royalty	%	R	6%	6%
Selling cost	%	S	0%	0%
Net Gold Price	$/oz.	Ng	1410	1,410
Met Recovery	%	Rec	92%	91%
Dilution	%	Dil	10%	10%
Ore Loss	%	Loss	2%	3%
Mining Cost - Contractor	$/t mined	MCC	2.95	3.52
Mining Cost - Owner’s	$/t mined	MCO	0.06	0.06
Mining Cost - GC	$/t mined	MCGC	0.07	0.14
Total Mining Cost	$/t mined	TMC	3.08	3.72
Strip Ratio	Waste/Ore	SR	13.62	3.46
G&A	$/t milled	G_A	7.7	8.8
Ore Crushing & hauling	$/t milled	rd	5.89	5.9
Mining	$/t milled	Cr	44.98	16.57
Process Plant	$/t milled	Cp	19	20.6
Maintenance/Engineering	$/t milled	Mp	0	0
Total Operating Costs	$		77.57	51.87
Full GradeCut-off	g/t		1.9	1.29
MarginalCut-off Grade	g/t		0.8	0.88

18^th September 2018 Page 181

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The Gounkoto undergroundcut-off grade calculations are presented in Table14-57.

Table14-57 Gounkoto Underground 2017Cut-Off Grade Calculations

Parameter	Unit	Gounkoto
Production
Life of Mine Production	Mt	13.82
Capital Cost	$M	112.42
Revenue Adjustment
NSR Royalty (Royal Gold: NSR)	of total oz produced	6%
Refining and Selling Expense	$/Ounce	-
Admin & Processing Costs
Metallurgical Recovery	%	94.2
Direct Processing	$/Ton	17.8
Tailings Disposal	$/Ton	-
Ore Haulage	$/Ton	-
Administration and Overhead	$/Ton	7.80
Total Other Cost	$/Ton	25.60
Mining Costs
Dilution	%	13.4%
Mining ore loss	%	4.0%
OPEX Development	$/Ton	6.62
OPEX Stoping	$/Ton	10.64
Backfill	$/Ton	14.31
Fixed Cost	$/Ton	14.34
Grade control	$/Ton	3.38
Total Mine Operating Cost	$/Ton	49.3
Mine Sustaining Capital	$/Ton	8.14
All in Cost	$/Ton	57.43
BreakevenIn-situCut-off Grade	g/t	2.30
Breakeven Mined Cut-off Grade	g/t	1.94
MarginalIn-situCut-off Grade	g/t	1.89
Marginal MinedCut-off Grade	g/t	1.60

14.16 Other Minor Satellites

At Loulo, there are additional small satellite deposits with historic Mineral Resources declared on them by Randgold. These deposits have not had exploration or Mineral Resource updates undertaken on them recently and constitute approximately 1% of the total contained gold ounces. Some of the minor satellite deposits have been partially mined at the surface for the soft oxide material. This was depleted from the Mineral Resources. These satellites are:

● P129

● P125L3

● P129QT

● Loulo 1

● Loulo 2 and L2L3Gap

● PQ10

These satellite deposits were sampled with a mixture of RC, diamond drilling, and trenching. Vertical sections were generated in Gemcom software on 12.5 m to 25 m spacing depending on the extents of the deposit and the drilling density. Hanging wall and footwall intersections were marked on the sections and the resultant strings were used to generate 3D mineralisation solids.

18^th September 2018 Page 182

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The mineralisation solids were clipped to the topographic surface and, where required, clipped to a basal surface that was generated at a distance below the bottom drill hole intercept. Weathering data obtained from drill sample logging was used to create surfaces tosub-domain the mineralisation solids into saprolite, transitional and fresh rock. Un-mineralised dykes were logged at some of the deposits, which were wireframes independently and used tocookie-cut the mineralisation wireframes.

Mineralised drill hole intercepts within the domain wireframes were composited, typically on1.0 m or 2.0 m lengths dependent on the Project. Composite sample datasets were reviewed to determine outliers fortop-cutting purposes. These analyses included histograms, log probability plots, and mean and variance plots. Where a dataset was observed to warranttop-cutting then this was applied to the composites. Not all datasets requiredtop-cutting.

Density measurements for each deposit were applied separately for the saprolite, transitional and fresh material. Where density measurements had not been collected for an individual deposit, the values obtained from another proximal deposit were applied.

Block models were generated using block sizes appropriate to the dimensions of the deposit and the drill hole spacing. Block models were interpolated primarily using OK, although some limited inverse distance weighting was used. Several passes of the search ellipsoids were used to ensure that every block was interpolated. Each subsequent run typically had a larger search volume and/or lower minimum sample number criteria.

The resultant block models were typically classified as Inferred Mineral Resources as a result of the low drill density and geological continuity. P129 and P125L3 contained sufficient drill densities to have kriging efficiencies and slope of regression values that would support an Indicated classification. In these areas, Indicated classification wireframes were generated to constrain the higher confidence material to improve continuity.

All minor satellite Mineral Resources were reported within a $1,500/oz pit shell at a 0.5 g/t Aucut-off.

The QP has confidence that these Mineral Resources were estimated with an appropriate level of accuracy and that the input data is valid and supports a Mineral Resource estimate. These minor satellite deposits do not have Ore Reserves declared on them. These deposits remain as suitable targets for expansion in future exploration.

14.17 Mineral Resources Reporting

The Mineral Resource estimates have been prepared according to the guidelines of the Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM)

18^th September 2018 Page 183

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI43-101 and there are no material differences.

The block model domains are filtered prior to reporting the Mineral Resource, to exclude anylow-grade areas that are unlikely to meet the criteria of having a reasonable prospect of eventual economic extraction.

All block model reports are generated using the “au_ok” field and are combined with other attributes flagged within the model depending on the purpose of the report.

The resource model was depleted using end of month 31^st December 2017 surveyed surfaces before reporting the 2017 declared Mineral Resources.

Thecut-off grade selected for limiting each of the Mineral Resources corresponds to the insitu marginalcut-off grade using a gold price of $1,500/oz.

For the open pit Mineral Resources, the pit shell selected for limiting of each of the Mineral Resources corresponds to a gold price of $1,500/oz. As a result of the optimisation process, this pit shell selection will result in the highest undiscounted net present value of the deposit, at $1,500/oz.

Underground panels were reviewed and those that were deemed as having a reasonable prospect of eventual economic extraction were included in the reported Mineral Resource.

The QP is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, fiscal, metallurgical, or other relevant factors, which could materially affect the Mineral Resource estimate.

Loulo

Table14-58 provides a detailed breakdown of the Loulo Mineral Resources including satellite deposits. The resource model was depleted using end of month 31^st December 2017 surveyed surfaces before reporting the 2017 declared Mineral Resources.

As of 31^st December 2017, the open pit Measured and Indicated Mineral Resources are 10.5 Mt at 2.8 g/t Au for 0.94 Moz.

The total underground Measured and Indicated Mineral are 43 Mt at 5.1g/t Au for 6.9 Moz.

Gounkoto

Table14-59 provides a detailed breakdown of the Gounkoto Mineral Resources including Faraba. The resource model was depleted using end of month 31^st December 2017 surveyed surfaces before reporting the 2017 declared Mineral Resources.

As of 31^st December 2017, the total Measured and Indicated Mineral Resources within the open pit and stockpiles are 25 Mt at 4.0 g/t Au giving 3.2 Mo Au.

18^th September 2018 Page 184

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The underground Measured and Indicated Mineral Resources are 3.0 Mt at 5.7 g/t Au giving 0.56 Moz Au.

In the QP’s opinion there are no environmental, permitting, legal, title, fiscal, socioeconomic, marketing, or other relevant factors which could materially impact the Gounkoto Mineral Resources.

18^th September 2018 Page 185

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-58 Loulo Total Declared Resources as of 31^st December 2017

Mineral Resource	Cut-off Grade   (g/t Au)	Measured			Indicated			Measured + Indicated			Inferred
		Tonnes (Mt)	Grade g/t	Gold (Moz)	Tonnes (Mt)	Grade g/t	Gold (Moz)	Tonnes (Mt)	Grade g/t	Gold (Moz)	Tonnes (Mt)	Grade   g/t	Gold (Moz)
Open Pit
Stockpiles	0.00	1.7	1.60	0.086				1.7	1.60	0.086
Loulo 3	0.73	0.31	3.82	0.038	2.5	4.30	0.34	2.8	4.25	0.38	0.77	5.2	0.13
Baboto	0.65	1.6	2.45	0.13	1.5	2.34	0.11	3.1	2.40	0.24	0.14	2.4	0.011
Gara West	0.75				2.6	2.35	0.20	2.6	2.35	0.20	0.32	2.6	0.026
P129	0.50				0.14	3.42	0.016	0.14	3.42	0.016	0.19	2.8	0.017
P125L3	0.50				0.16	2.51	0.013	0.16	2.51	0.013	0.045	2.4	0.0036
P129QT	0.50										0.11	2.6	0.0091
Loulo 1	0.50										0.45	2.4	0.035
Loulo 2 and L2L3Gap	0.50										0.45	1.8	0.027
PQ10	0.50										0.056	3.9	0.0071
Loulo OP Total		3.6	2.18	0.25	6.9	3.08	0.69	11	2.77	0.94	2.5	3.3	0.27
Underground
Gara	1.89	9.3	4.60	1.4	14	4.42	2.0	23	4.49	3.4	2.4	3.9	0.30
Yalea	2.04	7.3	5.48	1.3	12	6.06	2.3	19	5.84	3.6	7.4	4.2	1.0
Loulo UG Total		17	4.99	2.7	26	5.18	4.3	43	5.10	7.0	9.8	4.1	1.3
Open + Underground
Total Resources		20	4.49	2.9	33	4.73	5.0	53	4.64	7.9	12	3.9	1.6

Mineral Resources are reported on an 100% basis.

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

All Mineral Resource tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are those within a $1,500/oz pit shell.

Underground Mineral Resources are those below the $1,500/oz pit shell.

Mineral Resources for Gounkoto were generated by Mr Simon Bottoms, CGeol, an officer of the company and Qualified Person.

Numbers may not add due to rounding.

18^th September 2018 Page 186

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table14-59 Gounkoto Total Declared Resources as of 31^st December 2017.

Mineral Resource	Cut-Off     Grade (g/t)	Measured			Indicated			Measured + Indicated			Inferred
		Tonnes (Mt)	Grade g/t	Gold (Moz)	Tonnes (Mt)	Grade g/t	Gold (Moz)	Tonnes (Mt)	Grade g/t	Gold (Moz)	Tonnes (Mt)	Grade   g/t	Gold (Moz)
Open Pit
Stockpiles	0.00	1.8	1.96	0.11				1.8	1.96	0.11
Gounkoto	0.80	5.4	4.33	0.75	14	4.64	2.0	19	4.55	2.8	1.2	2.32	0.09
Faraba	0.88				4.3	2.14	0.29	4.3	2.14	0.29	0.2	2.35	0.01
Gounkoto OP Total		7.1	3.75	0.86	18	4.04	2.3	25	3.96	3.2	1.4	2.32	0.11
Underground
Gounkoto	2.00				3.0	5.74	0.56	3.0	5.74	0.56	2.6	3.51	0.29
Gounkoto UG Total					3.0	5.74	0.56	3.0	5.74	0.56	2.6	3.51	0.29
Open + Underground
Total Resources		7.1	3.75	0.86	21	4.28	2.9	28	4.15	3.7	4.0	3.09	0.40

The Mineral Resource estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Mineral Resources to CIM (2014) Standards, and there are no material differences.

Mineral Resources are reported on a 100% basis.

All Mineral Resources tabulations are reported inclusive of that material which is then modified to form Ore Reserves.

Open pit Mineral Resources are the insitu Mineral Resources falling within the $1,500/oz pit shells reported at a 0.80 g/t Aucut-off at Gounkoto and 0.88 g/t Aucut-off at Faraba.

Underground Mineral Resources are those insitu Mineral Resources within the ‘Jog Zone’ below the $1,500/oz pit shell reported at 2.0 g/t Aucut-off.

Mineral Resources for Gounkoto were generated by Simon Bottoms, CGeol, an officer of the company and Qualified Person.

18^th September 2018 Page 187

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.18 Comparison to Previous Models

Annual comparisons are completed to review the changes in the reported Mineral Resources due to model, depletion, andcut-off grade, where a 2017 model calculation using the previously declared resource is compared to the actual declared 2017 Mineral Resources.

Loulo

Yalea

Table14-60 shows the comparison between the 2017 declared Mineral Resource and the 2016 declared Mineral Resource.

Table14-60 Yalea 2017/2016 Mineral Resource Comparison

Change	Tonnes	Au Grade (g/t)	Au Ounces
2016 Declared Resource	26,789,377	5.62	4,836,579
2017 Depletion	1,502,269	8.18	395,020
Changes Due to Model	1,426,638	3.93	180,395
Changes Due toCut-off Grade	0	0.00	0
2017 Model Calculation	26,713,746	5.38	4,621,954
2017 EOY Projected Resources	26,713,746	5.38	4,621,954
Model Calculation vs Declared Resource Variance	0	0	0
2016 vs 2017 Resources	0%	-4%	-4%

The 180 koz Au increase can be broken up into:

● +432 koz gain in Yalea South from conversion drilling at higher grades defining inner edge of Yalea Southhigh-grade plunge and upgrading Inferred resources to Indicated.

● +90 koz gain in Yalea North from wider GC intersections than previously modelled.

● -100 koz loss at ‘Purple Patch’ because of relatively lower grade data added (added data mean of 7.64 g/t Au).

● -62.5 koz loss in ‘Purple Patch’ from density estimation and sub domaining of high- andlow-density domains

● -176 koz loss from Yalea North Central Deep Inferred with estimation and variogram update to prevent smearing of high-grade at depth.

18^th September 2018 Page 188

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gara

Table14-61 shows the comparison between the 2016 declared Mineral Resource and the 2017 declared Mineral Resource.

Table14-61 Gara 2017/2016 Mineral Resource Comparison

Change	Tonnes	Au Grade (g/t)		Au Ounces
2016 Declared Resource	25,695,030		4.29		3,546,201
2017 Depletion	1,181,286		3.75		142,315
Changes Due to the Model	1,153,457		6.86		254,478
Changes Due toCut-off Grade	0		-		0
2017 Model Calculation	25,667,201		4.43		3,658,365
2017 EOY Projected Resources	25,667,201		4.43		3,658,365
Model Calculation vs Declared Resource Variance	0.00%		0.00%		0.00%
2016 vs 2017 Resources	0.00%		3%		3%

The 180 koz gain Is the result of:

● +174 koz gain from GFSE conversion drilling (domain 9605) converting previously Inferred resources to Indicated.

● +80.5 koz gain in from wider GC intersections than previously modelled (including 36 koz gain within fold nose area of domain 8000).

Loulo 3

At Loulo 3 there were no model updates during 2017 or 2016, therefore the Mineral Resource remains the same as declared in 2016. Table14-62 shows the insitu resource comparison between the 2016 and 2017 declarations.

Table14-62 Loulo 3 2017/2016 Mineral Resource Comparison

Change	Tonnes	Au Grade (g/t)	Au Ounces
2016 Declared Resource	3,572,626	4.46		512,411
2017 Depletion	-	-		-
Changes Due to the Model	-	-		-
Changes Due toCut-off Grade	0	-		0
2017 Model Calculation	3,572,626	4.46		512,411
2017 EOY Projected Resources	3,572,626	4.46		512,411
Model Calculation vs Declared Resource Variance	0.00%	0.00%		0.00%
2016 vs 2017 Resources	0.00%	3%		3%

18^th September 2018 Page 189

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Baboto

Since the 2016 Mineral Resource declaration, Baboto North was transferred to Endeavour Mining Corporation. Table14-63 details shows the insitu resource comparison between 2016 and 2017 declaration.

Table14-63 Baboto 2017/2016 Mineral Resource Comparison

Change	Tonnes	Au Grade (g/t)	Au Ounces
2016 Declared Resource	5,825,615	2.13	399,421
2017 Depletion	0	-	0
Changes Due to the Removal of Baboto North	-1,924,096	1.62	-100,173
Changes Due to Model	-387,594	1.75	-21,865
Changes Due toCut-off Grade	97,113	0.69	2,165
Changes Due to Density	-334,013	2.52	-27,039
2017 Model Calculation	3,277,025	2.40	252,509
2017 EOY Projected Resources	3,277,025	2.40	252,613
Model Calculation vs Declared Resource Variance	0.00%	0.04%	0.04%
2016 vs 2017 Resources	-44%	12%	-37%

A total reduction of 146,808 koz Au is attributable to:

● -100 koz from sale of Baboto North to Endeavour Mining

● -22 koz due to changes in the modelling of the mineralisation which had the affect of increasing the grade slightly

● +2 koz increase reduction ofcut-off grade from 0.73 g/t Au to 0.65 g/t Au as result of processing cost decreases

● -27 koz from reduction in density values of saprolite and transition from 1.95 g/cm³ to 1.62 g/cm³ and 2.34 g/cm³ to 2.19 g/cm³ as a result of additional drilling to target density coverage

Gara West

There were no model updates or depletion during 2017, so the Mineral Resource remains the same as declared in 2016 (Table14-64).

Table14-64 Gara West 2017/2016 Mineral Resource Comparison

Change	Tonnes	Au Grade (g/t)	Au Ounces
2016 Declared Resource	2,954,753	2.37	225,608
2017 Depletion	-	-	-
Changes Due to the Model	-	-	-
Changes Due toCut-off Grade	0	-	0
2017 Model Calculation	2,954,753	2.37	225,608
2017 EOY Projected Resources	2,954,753	2.37	225,608
Model Calculation vs Declared Resource Variance	0.00%	0.00%	0.00%
2016 vs 2017 Resources	0.00%	3%	3%

18^th September 2018 Page 190

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

Gounkoto Open Pit

Table14-65 shows the changes from the 2016 Gounkoto open pit resources and the 2017 declared resources.

Table14-65 Gounkoto Open Pit 2017/2016 Mineral Resource Comparison

Change	Tonnes	Grade (g/t)	Au Ounces
Stockpile as of 31stDecember 2016	1,744,285	2.2	123,547
Gounkoto Main 2016 Insitu Resources	21,617,371	4.36	3,029,450
Faraba 2016 Insitu Resources	4,453,286	2.15	307,700
OP 2016 Declared Insitu Total Gounkoto Resources	27,814,941	3.87	3,460,697
Gounkoto Main Depletion (Jan 17 to Dec 17)	2,409,894	4.47	346,476
Total Model Change	930,793	5.94	177,648
Density Change	0	0	0
Economic Change	0	0	0
Pit Change	0	0	0
2017 Model Calculation	20,138,270	4.42	2,860,623
Stockpile as of 31stDecember 2017	1,761,603	1.96	111,212
Gounkoto Main 2017 Insitu Resources	20,177,638	4.41	2,863,320
Faraba 2017 Insitu Resources	4,453,286	2.15	307,700
OP 2017 Declared Insitu Gounkoto Resources	26,392,527	3.87	3,282,232
Model Calculation vs Declared Resource Variance	0.20%	-0.10%	0.10%
2016 vs 2017 Resources	-5.11%	-0.05%	-5.16%

A total reduction of 178 koz of gold is attributable to:

● 65% of the model change gain is from MZ1 due to the higher grades returned from 2017 infill grade control drilling in the dilation zone.

● MZ3 ounces gain as a result of the thinning of the mineralisation and higher than previously estimated grades returned by the deep diamond drilling targeting the high-grade zone seat at the bottom of the Super Pit.

● HW tonnes and ounces gain due to the GC drilling which widened HW1 as well as three new branches being added to the model

● MZ4 tonnes and ounces gain due to the extension of the domains over 100 m toward the south.

18^th September 2018 Page 191

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto Underground

Table14-66 outlines the comparison between the 2016 and 2017 Gounkoto underground estimates.

Table14-66 Gounkoto Underground 2017/2016 Mineral Resource Comparison

Change	Tonnes	Grade (g/t)	Au Ounces
Gounkoto UG Resource 2016	5,683,872	4.78	872,909
Depletion	0	0	0
Total Model Change	-47,500	10.44	-15,936
Density Change	0	0	0
Economic Change	0	0	0
Underground Area Change	-45,738	6.77	-9,959
2017 Model Calculation	5,590,635	4.71	847,014
UG 2017 Declared Insitu Gounkoto Resources	5,590,635	4.71	847,014
2017In-situ Mineral Resources vs 2016 DeclaredIn-situ Mineral Resources	-1.60%	-1.30%	-3.00%

The change between the estimates is a result of:

● Deepening of Life of Mine open pit which converted underground resources to open pit.

● Changes to MZ3 thickness from new data

Faraba

No model changes or depletion has been undertaken in 2017, therefore, the Mineral Resources are the same as declared in (Table14-67).

Table14-67 Faraba 2017/2016 Mineral Resource Comparison

Change	Tonnes	Grade (g/t)	Au Ounces
Faraba OP Resource 2016	4,453,286	2.15	307,700
Depletion	0	0	0
Total Model Change	0	0	0
Density Change	0	0	0
Cut-Off Grade Change	0	0	0
2017 Model Calculation	4,453,286	2.15	307,700
OP 2017 Declared Insitu Faraba Resources	4,453,286	2.15	307,700
2017In-situ Mineral Resources vs 2016 DeclaredIn-situ Mineral Resources	0%	0%	0%

18^th September 2018 Page 192

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.19 Reconciliation

Randgold has a standard weekly, end of month (EOM) and end of quarter (EQ) production measurement system that reports and provides reconciliation between grade control and the monthly mine production.

The Loulo and Gounkoto measurement system tracks daily, weekly, monthly, quarterly and year to date production Grade Control results versus the Plant. The system tracks both underground and open pit domains production against the block model. Summary reports are prepared weekly, monthly, and quarterly.

The reconciliation between GC call and Plant check out showed a moderate to poor reconciliation during 2017 mainly linked to the over-calling of underground-mined tons. Some of issues have been investigate and fixed (broken and crushed materials density fixed during the year), but the remaining issue that was resolved with the November 2017 Yalea resource model was the modelled insitu density at Yalea in PP.

Within the 2017 Mineral Resource estimate this was adjusted by utilising a single pass ID² interpolated density rather than a default value in the lower area of Yalea PP. After a single pass estimation, the mean density for each sub domain was then assigned to theun-estimated portion.

The resultant impact of the of the density estimate was a reduction of 264 kt for 62.5 koz metal loss relative to the previous single density assignment of 3.1 g/cm³ within the ‘Purple Patch’, but with a directly associated 5% improvement on the mine call factor oz reconciliation for 2017 (Table14-68).

Table14-68 Loulo Gounkoto MCF 2017 EOY Reconciliation (Old assign density LEFT and new estimated density + domain assign RIGHT)

Dept	Recon Ore Mine, Stockpiles and Plant Out	Assigned Density			Estimated Density
		Tons	Grade	Ounces	Tons	Grade	Ounces
GC	Mine	5,027,562	5.24	846,862	5,027,562	5.24	846,862
GC	Stockpile Change	-64,908	9.60	-20,025	-64,908	9.60	-20,025
GC	GC Actual Feed	5,117,531	5.22	859,229	5,117,531	5.22	859,229
GC	Scats Stock Change	-54,323	2.12	-3,711	-54,323	2.12	-3,711
Plant	Cone Change	-4,806	18.98	-2,932	-4,806	18.98	-2,932
GC	GC Call	5,176,661	5.19	863,645	5,176,661	5.19	863,645
GC	Plant Check Out	4,918,317	4.96	784,877	4,918,317	4.96	784,877
GC vs Plant	MCF (%) GC Call vs Plant Check In	95	94	89	96	98	95
GC vs Plant	MCF (%) GC Call vs Plant Check Out	95	96	91	96	100	96

If the density at PP had not been adjusted this would have been a material impact on the 2018 mine plan as 63% of underground stopes planned fall in the PP, and most of those are in the fringe area where the density is lower than that of the average for PP (3.1 g/cm³).

18^th September 2018 Page 193

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

14.20 External Audits

Loulo

An external audit was completed in January 2015 of the Yalea and Gara Mineral Resources by QG.

QG concluded that following their site visit and reviews of the geology and Mineral Resource that:

QG is confident that the 2014 Yalea and Gara Mineral Resources are free of material error. Classification of some of the Indicated Resources at Yalea and particularly Gara is at the limit of Indicated and requires further drilling without necessarily changing/improving the classification.

There are no issues that QG consider need immediate attention. Any recommendations made are more for long term improvement in processes and understanding. QG notes that Randgold has built a strong team to drive the resource estimation process with good, capable practitioners on the ground.

QG considered that both the broad and local geology at both Yalea and Gara is well understood and that Randgold has a strong and capable team on the ground.

QG outlined several areas where incremental improvements of the procedures that could be undertaken. QG’s principal concern was around the extent of drilling within the Indicated material. Since this review was undertaken, significant drilling has been undertaken which has addressed this concern. Other minor improvements included:

● Use of a certified weight for calibration during bulk density measurements.

● Logging directly into a digital system on rugged computers.

● Reduce the number of different CRM’s being used at any one time.

● Review QA/QC on periods longer than nine months.

● Undertake a trial estimate where density values are estimated into the block model.

● Applying dynamic anisotropy to estimations to reduce the complexity of orientation domains applied.

All the recommendations made by QG have either been completed, are in the process of being updated, or have been tested and reviewed by Randgold.

18^th September 2018 Page 194

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto

An external audit completed in February 2015 by QG. QG concluded that the“Gounkoto Mineral Resource is free of error”. QG principle concern with Gounkoto was the geological complexity of the deposit, compared to Yalea and Gara, which warranted additional drilling prior to the development of Gounkoto Underground. QG did note that the broad Gounkoto geology was well understood.

The results concluded that the defining of alteration, veining, mineralisation, and package boundaries was well understood. They note that the Gounkoto geologists have used a conservative approach during wireframing by using alteration as a means of limiting domain boundaries. Although this can result in the exclusion of somelow-grade, inter-alteration assays, this can be classed as a positive methodology if it is mined underground as ‘decisions will more often than not be made on observation not sampling’.

Since this review has taken place, significant additional drilling has taken place to improve confidence in the geological models and the Gounkoto Life of Mine open pit shell has been extended significantly as part of the ‘Super Pit’, which has converted underground resources into open pit resources. In 2016, an independentpre-feasibility study (PFS) was undertaken by Piran Mining Pty Ltd to evaluate the Gounkoto Underground. This PFS deemed that underground mining could be profitable to 540 m below the surface at a $1,000/oz gold price.

Other incremental changes to the Randgold procedures included:

● Use of calibrated weights for density measurements

● Reduction in quantity of different CRMs being used

● Applying dynamic anisotropy to estimations to reduce the complexity of orientation domains applied.

All the recommendations made by QG have either been completed, are in the process of being updated, or have been tested and reviewed by Randgold.

14.21 Discussion

The QP is of the opinion that the 2017 Mineral Resource estimates are free of material error and are a true reflection of the geology and mineralisation that has been observed at both Loulo and Gounkoto.

Randgold plant to implement dynamic anisotropy in folded bodies during the 2018 Mineral Resource updates to reduce the complexity of sub-domaining and improve the quality of local estimation.

Future model revisions at Yalea for year end 2018 Mineral Resource updates will include changes within Yalea South Transfer zone with additional drilling that is planned for 2018.

18^th September 2018 Page 195

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

A geological model review and targeted drilling is planned for 2018 for the purposes of defining additional open pit and underground resources at Loulo 3.

Drilling is planned at Yalea Central Deep zone to test the Inferred material with the purpose of increasing the confidence of the estimate within this zone.

In 2018, Yalea South infill grade control drilling is required to target Indicated resources for the purpose of upgrading to Measured Mineral Resources prior to first production which is scheduled in Q4 2018.

Relative Accuracy / Confidence of the 2017 Mineral Resource Estimate

The QP is of the opinion that the input data for the 2017 model updates is accurate and supports the declaration of a Mineral Resource estimate. The resource estimate has followed industry standard practices for collecting, validating, and estimating data.

The application of thorough exploratory data analysis and quantitative neighbourhood kriging analysis gives a high confidence in the Measured and Indicated material in the model. The Inferred material by nature has a relative low level of accuracy, but a high geological confidence to exist.

As Loulo and Gounkoto are operating mines, grade control and advanced grade control drilling have provided a much higher degree of accuracy in the model and the ability to reconcile mine data to compare with the block model. Year on year model comparisons show a very good reconciliation between the theoretical Mineral Resources and the declared Mineral Resources.

External audits completed by QG consultants in early 2015 documented that, at the time, there was no material concerns with the Mineral Resource estimates of either Loulo or Gounkoto. Since these audits were completed, incremental and documented updates have occurred annually. Minor concerns that QG outlines with the complexity of the Gounkoto mineralisation have been addressed with additional drilling, model updates and apre-feasibility study.

Additional grade control drilling in 2017 has confirmed previous model concepts and provided confidence in the Mineral Resource estimate.

18^th September 2018 Page 196

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

15 Ore Reserve Estimate

The Ore Reserve estimates have been prepared according to the guidelines of the Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves standards and guidelines published and maintained by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy and the Australian Institute of Geoscientists and Minerals Council of Australia (the JORC (2012) Code). Randgold has reconciled the Mineral Resources and Ore Reserves to Canadian Institute of Mining, Metallurgy and Petroleum (CIM) 2014 Definition Standards for Mineral Resources and Mineral Reserves dated May 10, 2014 (CIM (2014) Standards) as incorporated with NI43-101 and there are no material differences.

Definitions for reserve categories used in this report are consistent with those defined by the CIM (2014) Standards and adopted by NI43-101. In the CIM classification, a Mineral Reserve is defined as “those parts of Mineral Resources which, after the application of all mining factors, result in an estimated tonnage and grade which, in the opinion of the QP(s) making the estimates, is the basis of an economically viable project after taking account of all relevant Modifying Factors. Mineral Reserves are inclusive of diluting material that will be mined in conjunction with the Mineral Reserves and delivered to the treatment plant or equivalent facility”. Mineral Reserves are classified into Proved and Probable categories.

The term ‘Ore Reserves’ as used in this report is synonymous with ‘Mineral Reserve’ as defined by the CIM.

The QP has performed an independent verification of the block model tonnes and grade, and in their opinion, the process has been carried out to industry standards

The QP is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, fiscal, infrastructure, permitting, that could materially affect the Ore Reserve estimate.

Loulo

The Loulo gold mine consists of two main deposits, Gara and Yalea, and multiple satellites deposits. Both Gara and Yalea are underground mining operations and their Ore Reserves have been updated with the 2017 additional work in drilling and mapping providing further information to improve the geological model for resources estimation.

In addition, the satellite deposit of Baboto has been updated with a new model for the 2017 declaration. No new data has been added to the satellite deposits of Loulo 3 and Gara West since the previous model update, however the models have been updated with the current estimation processes used for Yalea and Gara.

Both the Gara and Yalea models incorporate data from the open pit and underground grade control infill drilling as well as exploration diamond holes. Thecut-off date for data used to create the models is 20^th October 2017 for Yalea and 20^th November 2017 for Gara. For the Baboto

18^th September 2018 Page 197

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

satellite, data was incorporated until end of 31^st May 2017. No drilling that would make a material impact on the Ore Reserve was completed between thecut-off date and the effective date of the Mineral Resource (31^st December 2017).

For Loulo, all Mineral Resources model updates since 2014 were prepared using Maptek Vulcan software. Some earlier models of the minor deposits were created in Gemcom and have not been updated yet. Table15-1 summarises the Loulo Ore Reserve estimate as of 31^st December 2017.

The total Proved and Probable Ore Reserves estimated for Loulo as of 31^st December 2017 are 36 Mt at 4.50 g/t Au containing 5.2 Moz of gold. The Ore Reserves consist of surface stockpiles, Loulo open pit (OP) and underground (UG) Ore Reserves.

Table15-1 Loulo Gold Mine Ore Reserve Statement as of 31^st December 2017

Source	Ore Reserve	Tonnes   (Mt)	Grade g/t	Gold  (Moz)	*Attributable Gold (Moz)*
Stockpiles	Proved	1.7	1.60	0.086	0.068
	Proved	1.5	2.43	0.12	0.093
Open Pit	Probable	3.9	3.87	0.48	0.39
	Proved + Probable	5.4	3.47	0.60	0.48
	Proved	8.8	4.97	1.4	1.1
Underground	Probable	20	4.82	3.1	2.5
	Proved + Probable	29	4.87	4.5	3.6
	Proved	12	4.18	1.6	1.3
Total	Probable	24	4.67	3.6	2.9
	Proved + Probable	36	4.51	5.2	4.1

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in the Loulo Gold Mine. Ore Reserves are reported on a 100% and attributable basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz and an averagecut-off grade of 1.1 g/t Au including dilution and ore loss factors.

Underground Ore Reserves are reported at a gold price of $1,000/oz and acut-off grade of 2.69 g/t Au for Yalea and 2.4 g/t Au for Gara including dilution and ore loss factors.

Open pit and underground Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

Gounkoto

Three primary sources of ore contribute to the 2017 Gounkoto Ore Reserve. Gounkoto open pit and underground, and Faraba open pit. The Gounkoto open pit and underground models were updated in 2017 as a result of depletion at Gounkoto open pit along with new drilling and model changes for both open pit and underground operations. The Faraba model remains the same as 2016 as no additional work has been undertaken on the deposit.

Thecut-off date for data used in the Gounkoto update is 14^th July 2017. All material drilling was included in the Faraba reserve.

All Mineral Resource models for Gounkoto have been created using Maptek Vulcan. Table15-2 summarises the Gounkoto Ore Reserve estimate as of 31^st December 2017.

The total Proved and Probable Ore Reserves for Gounkoto as of 31^st December 2017 are estimated to be 20 Mt at 4.58 g/t Au containing 3.0 Moz gold.

18^th September 2018 Page 198

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-2 Gounkoto Gold Mine Ore Reserve Statement as of 31^st December 2017

Source	Ore Reserve	Tonnes (Mt)	Grade g/t	Gold (Moz)	Attributable Gold (Moz)*
Stockpiles	Proved	1.8	1.96	0.11	0.099
	Proved	4.4	4.75	0.66	0.53
Open Pit	Probable	12	4.63	1.8	1.4
	Proved + Probable	16	4.66	2.4	1.9
	Proved	-	-	-	-
Underground	Probable	2.2	6.09	0.42	0.34
	Proved + Probable	2.2	6.09	0.42	0.34
	Proved	6.1	3.95	0.78	0.62
Total	Probable	14	4.85	2.2	1.7
	Proved + Probable	20	4.58	3.0	2.4

*Attributable gold (Moz) refers to the quantity attributable to Randgold based on Randgold’s 80% interest in the Gounkoto Gold Mine. Ore Reserves are reported on a 100% and attributable basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz and an averagecut-off grade of 1.1 g/t Au including dilution and ore loss factors.

Underground Ore Reserves are reported at a gold price of $1,000/oz and acut-off grade 3.0 g/t Au including dilution and ore loss factors.

Open pit and underground Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

15.1 Introduction

For reserve and operational discussion, future and current operations are generally grouped under the following structure:

● Open pits

^○ Loulo (no current production)

• Gara West

• Loulo 3

• Baboto (scheduled for 2018)

^○ Gounkoto

• Gounkoto (operating)

• Faraba

● Underground

^○ Loulo

• Yalea (operating)

• Gara (operating)

^○ Gounkoto

• Gounkoto(pre-feasibility level deposit)

Some inputs were shared across all the operations during the preparation of the Reserve estimates. Ore Reserves were based on the development of appropriately detailed and engineered LOM plans. All design and scheduling work have been undertaken to an appropriate level of detail by experienced engineers using appropriate mine planning software. The planning process incorporated appropriate modifying factors and the use ofcut-off grades and other

18^th September 2018 Page 199

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

technical-economic investigations. Further, in presenting the Ore Reserve statements and associated sensitivities the following applies:

● The Measured and Indicated Mineral Resources are inclusive of Ore Reserves.

● All Ore Reserves are estimated as of 31^st December 2017.

● All Ore Reserves are estimated at a gold price of $1,000/oz.

● All Ore Reserves are estimated in terms ofRun-of-Mine (RoM) grades and tonnage as delivered to the metallurgical processing facilities and are fully diluted.

● Ore Reserves include only Measured and Indicated Mineral Resources with appropriate modifying factors applied and are used in the LOM plan.

● All references to Mineral Resources and Ore Reserves are stated in accordance with the JORC (2012) Code and reconciled to CIM (2014) Standards adopted by NI43-101.

● All ounces referred to in this report are Troy ounces, equivalent to 31.10348 grammes.

● Ore Reserves are stated as total Life of Mine tonnes and gold content as well as Life of Mine gold content that is attributable to Randgold due to the percentage of the operation or Project that Randgold owns.

The JORC (2012) Code allows a limitednon-material quantity of Inferred Resources to be included in an estimate of Ore Reserves or to support a cash flow, however, a small portion (0.8% of the scheduled tonnes) of the Gounkoto Project are classified as Inferred and have been included in the cash flow projection. This is due to the application of practical mining shapes to the resource model during evaluation. This proportion is deemed small enough to be acceptable to be included in the overall Ore Reserve report.

The location of the operations and deposits is shown in Figure5-1 and Figure5-2.

A summary of the Ore Reserves is given in Sections 15.2 to15.5, with a more detailed description of the mine design and modifying factors in Sections15.6-15.9.

15.2 Loulo Open Pit Ore Reserve

Surface sources at Loulo comprise the Baboto Pit, Loulo 3 Pit, the Gara West Pit, and some stockpiles.

Some of the Baboto pit was mined in 2018, however the rest of Baboto pit and the other pits will be mined from 2024, after the larger Gounkoto pits start to deplete.

A decision was made in 2017 to sell off the northern portion of the Baboto deposit. This has resulted in a reduction of 76 koz Au to the declared Ore Reserve.

As stockpiles are managed and known, they have been classified as Measured Mineral Resources. The stockpiles include material from all the current operations.

The Loulo open pit Ore Reserve estimate is presented in Table15-3. Further details on the stockpile Ore Reserve is presented in Table15-4.

18^th September 2018 Page 200

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-3 Loulo Open Pit Ore Reserve as of 31^st December 2017

Source	Ore Reserve	Tonnes (Mt)	Gold (g/t)	Gold (Moz)
Stockpiles	Proved	1.7	1.60	0.086
	Proved	-	-	-
Gara West Pit	Probable	0.86	2.62	0.073
	Proved + Probable	0.86	2.62	0.073
	Proved	-	-	-
Loulo 3 Pit	Probable	2.2	4.90	0.35
	Proved + Probable	2.2	4.90	0.35
	Proved	1.5	2.43	0.12
Baboto Pit	Probable	0.82	2.40	0.064
	Proved + Probable	2.3	2.40	0.18
	Proved	3.2	1.98	0.21
Total	Probable	3.9	3.86	0.48
	Proved + Probable	7.0	3.02	0.69

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz and an averagecut-off grade of 1.1 g/t Au including dilution and ore loss factors.

Open pit Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

Table15-4 Loulo Surface Stockpile Ore Reserve as of 31^st December 2017

Location	Stockpiles		Cut-Off (g/t)	Tonnes (kt)	Gold (g/t)	Gold (koz)
Rom Pad	Transition	Loulo 3	>1.36 - 3.23	491	1.59	25
		Yalea OC	>1.09 - 2.60	347	1.48	16
		Gara OC	>1.04 - 2.07	-	-	-
	Fresh	Yalea UG	>2.78	0.51	8.46	0.14
		Gara UG	>2.56	0.072	2.99	0.0070
		Big Blocks (Gara/Yalea	>1.09 - 2.60	2.3	2.13	0.16
	Transition	Yalea OC	>2.60	2.7	3.13	0.28
Rehandle	Fresh	Yalea OC	>1.09 - 2.60	588	1.61	30.35
		Gara OC	>1.04 - 2.07	182	1.50	8.80
		Big Blocks (Gara/Yalea	>1.09 - 2.60	13	2.27	0.93
Total Ore Stockpiled				1,626	1.57	880.60
Plant Cone	Fresh	Yalea UG	>2.78	-	-	-
		Gara UG	>2.56	1.2	3.50	0.14
	Strategic Cone	Mixed Ore		24	2.71	2.10
	Old Cone	Mixed Ore		2.5	2.87	0.23
	3rd Mill Cone	Mixed Ore		7.3	3.52	0.83
Total Plant Cone				35.2	2.92	35.2
Total Stockpile				1,661	1.60	85

Stockpiles are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Stockpiles were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

18^th September 2018 Page 201

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

15.3 Gounkoto Open Pit Ore Reserve

Surface sources at Gounkoto comprise the Gounkoto pit, also referred to as the ‘Super Pit’, which provides the bulk of the ore, and the Faraba pit. The Faraba pit will be mined from 2022 as the Gounkoto pit production winds down.

The Gounkoto open pit Ore Reserve estimate is presented in Table15-5.

Table15-5 Gounkoto Open Pit Ore Reserve as of 31^st December 2017    

Source	Ore Reserve	Tonnes   (Mt)	Gold g/t	Gold   (Moz)
Stockpiles	Proved	1.8	1.96	0.11
Gounkoto Pit	Proved	4.4	4.75	0.66
	Probable	9.3	5.19	1.6
	Proved + Probable	13	5.05	2.2
Faraba Pit	Proved	-	-	-
	Probable	2.5	2.51	0.20
	Proved + Probable	2.5	2.51	0.20
Total	Proved	6.1	3.95	0.78
	Probable	12	4.63	1.8
	Proved + Probable	18	4.40	2.5

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Open pit Ore Reserves are reported at a gold price of $1,000/oz and an averagecut-off grade of 1.1 g/t Au including dilution and ore loss factors.

Open pit Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

15.4 Loulo Underground Ore Reserve

Underground sources at Loulo comprise the Yalea and Gara mines. Both of these are currently being mined. No stockpiles have been included as these were included in the Loulo surface Ore Reserves.

The conversion percentage of Measured and Indicated Mineral Resource to Ore Reserve is Yalea 67% and Gara 61%. The Loulo underground Ore Reserve estimate is presented in Table15-6.

18^th September 2018 Page 202

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-6 Loulo Underground Ore Reserves as of 31^st December 2017

Source	Ore Reserve	Tonnes (Mt)	Gold (g/t)	Gold (Moz)
Yalea Underground	Proved	3.4	6.40	0.71
	Probable	10	5.43	1.8
	Proved & Probable	13	5.68	2.5
Gara Underground	Proved	5.4	4.05	0.70
	Probable	10	4.22	1.4
	Proved & Probable	15	4.16	2.1
Total Loulo Underground	Proved	8.8	4.97	1.4
	Probable	20	4.83	3.1
	Proved & Probable	29	4.87	4.5

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Underground Ore Reserves are reported at a gold price of $1,000/oz and acut-off grade of 2.69 g/t Au for Yalea and 2.4 g/t Au for Gara including dilution and ore loss factors.

Underground Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

15.5 Gounkoto Underground Ore Reserve

The Gounkoto Underground project is the only source of underground ore for the Gounkoto underground Ore Reserve estimate, presented by category, in Table15-7.

Table15-7 Gounkoto Underground Project – Total Ore Reserve by Category as of 31^st December 2017

Source	Ore Reserve	Tonnes (Mt)	Gold (g/t)	Gold (Moz)
Total Gounkoto     Underground	Proved	-	-	-
	Probable	2.2	6.09	0.42
	Proved &   Probable	2.2	6.09	0.42

Ore Reserves are reported on a 100% basis.

The Ore Reserve estimate has been prepared according to JORC (2012) Code. Randgold has reconciled the Ore Reserves to CIM (2014) Standards, and there are no material differences.

Underground Ore Reserves are reported at a gold price of $1,000/oz and acut-off grade 3.0 g/t Au including dilution and ore loss factors.

Underground Ore Reserves were estimated by Mr. Derek Holm, FSAIMM, an external consultant and Qualified Person.

Numbers may not add due to rounding.

15.6 Reserves Estimation Process

The same Ore Reserve estimation process was followed for all the open pits. Where relevant, details have been given for the separate deposits.

The Loulo underground Ore Reserve estimate is based on two operating mines, Yalea and Gara. As such, the operating parameters of those deposits are well known. The Gounkoto underground Ore Reserve estimate is based on aPre-feasibility study with a lower level of confidence. While the design method and parameters for future underground mining are similar, due to the lower level of confidence, these have been tabled separately.

18^th September 2018 Page 203

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

15.7 Geotechnical Parameters

Loulo Open Pit Operations

A geotechnical assessment was completed by Dr. Peter Gash (Open Pit Slope Design for the Baboto Pits, MineNet, April 2017) on the Baboto deposit in 2017. The results of this assessment have been used to update the Baboto pit design.

A summary of the parameters applied to the Baboto, Loulo 3, and Gara West pit designs is shown in Table15-8.

There is no geotechnical reason to reduce the bench heights to 5 m in soils. This is driven by the contractor’s preferences.

18^th September 2018 Page 204

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-8 Baboto – Loulo 3 - Gara West - Slope Parameters

Area/Sector	Inter Ramp (°) (crest- crest)			Bench Height (m)			Bench Angle (°)			Berm (m)
	Baboto	Loulo 3	Gara West	Baboto	Loulo 3	Gara West	Baboto	Loulo 3	Gara West	Baboto	Loulo 3	Gara West
Soil Slope - Bench Configuration
Bench stacks for wall up to 40 m (depth <40 m)	44	42	46	10	5	5	60	65	60	4.5	3.2	2.0
Bench stacks for wall greater 40 m (depth>40 m)	40	40	40	10	5	5	60	65	60	6.0	3.5	3.0
Saprock Slope - Single Bench Configuration
Footwall	47.4	N/A	50	10	N/A	10	75	N/A	75	6.5	N/A	5.5
Hanging Wall	47.4	N/A	52	10	N/A	10	75	N/A	75	6.5	N/A	5.0
Rock Slope - Single Bench Configuration
Footwall	50.7	48.7	50	10	10	10	75	65	75	5.5	3.0	5.5
Hanging Wall	52.5	49.0	52	10	10	10	75	70	75	5.0	2.9	5.0
Rock Slope - Twin Bench Configuration
Footwall	52.5	N/A	N/A	10	N/A	N/A	75	N/A	N/A	8.0 + 2.0 offset	N/A	N/A
Hanging Wall	55.9	N/A	N/A	10	N/A	N/A	80	N/A	N/A	8.0 + 2.0 offset	N/A	N/A

18^th September 2018 Page 205

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto Open Pit Operations

The strength of the rock mass at Gounkoto improves moving away from the orebody rocks and into the walls. The weathering profile is deep in the NE, withtop-of-fresh rock averaging approximately 60 m up to a maximum of 80 m depth below original ground surface.

Slopes in soils are controlled by mass stability conditions and are sensitive to both height and water pressure. The design parameters are therefore qualified by slope height and dewatering measures.

Based on experience to date a twin benching configuration for rock and soil slopes was applied, with the following geometry in each sector (Table15-9).

Table15-9 Geotechnical Slope Requirements

Area/Sector	Soil/Rock		Inter Ramp (°) (crest- crest)		Bench Height (m)		Bench Angle (°)		Offset (m)		Berm (m)
Soil Slope - Bench Configuration
Bench stacks for wall up to 40 m (depth <40 m)		All		44		10		60		NA		4.5
Bench stacks for wall greater 40 m (depth>40 m)		All		40		10		60		NA		6
Rock Slope - Single Bench Configuration
Footwall + Side wall (North and South)		All		50.7		10		75		NA		5.5
Hanging Wall (North and South)		All		54		10		80		NA		5.5
Rock Slope - Twin Bench Configuration
Footwall + Sidewall - North		All		52.5		10		75		2		8
Footwall + Sidewall - South		All		51.6		10		75		2		8.5
Hanging Wall (North and South)		All		55.9		10		80		2		8

Slope Monitoring

A radar system (IBIS Rover mobile unit) was purchased in 2017 and installed and provides a complete support for the geotechnical interpretation of slope deformation behaviour.

Dewatering for Slope Stability Improvement

A total of 50 boreholes at an average depth of 100 m will be installed for the expansion of the Super Pit.Six-inch submersible pumps will be installed in two thirds of these boreholes with the remainder acting as monitoring boreholes to keep track of the prevailing Ru (ratio of water pressure to soil pressure). This needs to be below 0.5 as a stability condition for saprolite/soil and/or weak rock slopes. In seasons where the Ru may rise above the allowable 0.5, pumps will be installed on the monitoring boreholes to keep the Ru within stable limits.

Most of the rock slope will be developed below the water table and a network/series ofsub-horizontal drains (weep holes) will continue be used to depressurise the wall and improve the stability of the rock slopes. Thesesub-horizontal holes will be drilled at three holes per fan and each hole will be 50 m into the wall. Each fan will be planned targeting water bearing structures. Thesub-horizontal holes will be free draining and pumps will not be installed in such holes.

18^th September 2018 Page 206

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo Underground Operations

A geotechnical assessment was undertaken by SRK as part of the LouloPre-Feasibility and Feasibility Studies. Since commencement of underground mining, geotechnical assessments have been undertaken by Middindi Consulting Pty Ltd (Middindi) in 2011 (Geotechnical Report on the Stability and Mine Design of Yalea Underground), and more recently by Dempers and Seymour in 2013 (Loulo Underground Geotechnical Review)

Since 2013, a significant amount of geotechnical work has been undertaken for both Yalea and Gara:

● Measurement ofpre-mining rock stress.

● Construction of geotechnical rock mass models.

● Modelling (Map3D) of mining induced stresses of proposed mining sequences.

● Creation of stable span stability curves based on analysis of local data.

● Building of the capability of the site geotechnical team.

● Development of a Dilution Rating System (DRS) at Gara only to improve prediction stope geotechnical conditions.

This work has enabled the refinement of the mine designs. The rock mass models have provided a better understanding of the variability in hanging wall ground conditions at Yalea and Gara. The stoping designs in the mine design have maximum hanging wall or footwall exposure of a hydraulic radius of 10 m (30 m vertical, 50 m along strike with a 65° dip). However, the majority of stoping will use an underhand (top down) with paste fill mining method that has only single height stope exposures with hydraulic radius 8.6 m (25 m vertical). This design is conservative from a hanging wall span perspective, however from an overall mining perspective it has economic benefits. The basic geotechnical parameters of the operations are given in Table15-10.

Table15-10 Basic Geotechnical Parameters for Yalea and Gara

Source	Planned Height (m)	Planned Maximum Width (m)	Maximum Length (m)	Hydraulic radius (m)
Yalea
Stopes	25	5 to 30	30 to 60	7 to 9
Pillars	20 to 25	5 to 30	10
Gara
Stopes	25	2 to 20	30	7
Pillars	20 to 25	5 to 20	10

Gounkoto Underground Operations

Mining Rock Mass Models have been updated for Gounkoto based on the updated structures, lithological wireframes, and the latest drilling data. This current model version is Gounkoto V5 dated September 2016.

D&S was commissioned by Randgold to update the 3D model of geotechnically significant structures and the Mining Rock Mass Model (MRMM) for the Gounkoto Project.

18^th September 2018 Page 207

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The modelled structures include Fault Gouge, Fault West Dipping, the faults intersecting MZ3 orebody, and the faults in footwall of MZ1 orebody (excluding the Fault Breccia which is not geotechnically significant) have been modified and updated with the geotechnically significant intercepts and used in updating the MRMM. Two wireframes of structure in the footwall of MZ1 do not have any geotechnically significant intercepts were excluded from the MRMM update.

The rock quality and stress to strength ratios expected to be encountered at Gounkoto are predominantly in the moderately fractured, low to intermediate stress regions of the tunnel instability chart, with limited areas of highly fractured conditions. The basic geotechnical parameters of the project are given in Table15-11.

Table15-11 Basic Geotechnical Parameters for Gounkoto

Source	Planned   Height (m)	Planned Maximum Width (m)	Maximum   Length (m)	Hydraulic   radius (m)
Gounkoto
Stopes	23	2.5 to 41	25	6
Pillars	18	2.1 to 41	25

The Gounkoto main pit ramp and decline will serve as the main access for a planned underground mine life of six years. However, it is possible that the mine and main infrastructure life may be considerably greater, so geotechnical design standards were increased to cover a longer potential life of over 10 years.

Internationally accepted development surface support standards will be used to minimise the exposure of personnel and equipment to rockfall hazards. Therefore, all support recommendations will comply with the Mines Occupational Safety and Health Advisory Board (MOSHAB, Australia1999) Code of Practice for Surface Rock Support. This code requires surface support (retainment) of any development backs and walls that are greater than 3.5 m above the floor, the height above which check scaling by hand is considered to be impractical.

Ground support recommendations have considered:

● The expected stress regime, rock quality and expected modes of tunnel instability.

● Empirical support design methods.

● Potential wedge instability.

● Excavation size and duty.

15.8 Parameters Affected by Groundwater Control

Loulo Open Pit

The pit slope calculations for all Loulo Open Pits are based on the assumption that the pit slopes will be dry. With this in mind, ground dewatering is planned and implemented well before the start of mining. The only active mining open pit currently at Loulo is the Baboto pit.

18^th September 2018 Page 208

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

At Baboto over much of the planned perimeter, the regolith is relatively shallow and the slope is expected to be stable. However, the North sector of the South Pit has deep penetrative saprolitisation and this sector will require advance dewatering.

The zones for dewatering installations are shown in Figure15-1.

Figure15-1 Sketch of Baboto Dewatering Arrangement

Similar ground water control plans will be developed prior to the mining of the Gara and Loulo 3 pits.

Gounkoto Open Pit

The motivation for an active dewatering strategy is to improve slope stability principally in the regolith slopes, which in turn reflects in a significant increase in the allowable angle. Groundwater control is an absolutepre-condition for adopting the slope design above the fresh rock. The strategy is multi-faceted:

● Borehole wells (pumps) – in the regolith.

● Excavation rate – controlling the rate of deepening in saprolite.

● Horizontal drains – in rock at depth.

● Sump strategy - planning by season, cross-fall, capacity.

● Surface works – diversions,run-off control.

The impact of ground water on strong rock slopes is secondary in most cases and is generally manageable byin-pit horizontal drains. The motivation for dewatering wells is the regolith. For Baboto, systematic horizontal drains will not be needed, unless an exceptional water ingress problem is encountered.

The principal impact of effective dewatering is an increase in the allowable slope angle of up to 15°(I-R). As a consequence of recognising and implementing this strategy, the Randgold pits have steeper regolith slopes than other equivalent pits in the region.

Pumps will be installed in advance of excavation. Drawdown takes time and experience shows that installation around six months in advance of digging is effective.

18^th September 2018 Page 209

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo Underground

A groundwater monitoring program was implemented at Loulo prior to the commencement of mining operations. Since 2015 Artois Consulting have been working on improving the understanding of hydrogeology and water management at the Loulo mines. In 2017 the main focus of this work was on the Yalea south proposed SURF mining area where weathering extends approximately 300 m below surface.

The weathered area needs to be dewatered to enable the stopes in and near the weathered zone to be mined safely. Piezometric sensors have been installed in bore holes from surface, dewatering holes have been drilled from underground and dewatering has commenced.

This area was saturated to +25 MASL (metres above mean seal level; 135 m below surface). Dewatering has drawn down the ground water level in one hole on the footwall side of the ore zone to-90 MASL (250 m below surface). Further dewatering holes through the weathered zone are required to draw down the water in the hanging wall side of the ore zone.

Gounkoto Underground

Groundwater conditions are likely to be similar to those encountered in the Gara and Yalea mines in that they are not expected to be adverse in terms of either:

● Water quality, with fairly benign pH, TDS and contained salts.

● water ingress frequency, volume, and pressure.

No mine design changes were required to account for groundwater management.

A permanent pump station has been designed at the bottom of the underground workings. It is located near the end of the decline and will pump to the elevation of the portal and discharge into the pit workings. Pumped water will be routed via pipes situated in service drives located off the decline. This pumping arrangement is similar to that of Yalea and Gara and will be further detailed closer to the start of production.

15.9 Dilution and Mining Recovery

Open Pit Operations

For the Loulo and Gounkoto open pits, a dilution figure of 10% at zero grade was applied, with an ore loss of 2% (3% for Faraba). In some areas of the pit the dilution does contain grade, but the operation has reconciled their production to an equivalent value based on zero grade.

Reconciliations are completed monthly and where required are adjusted to suit the mining conditions. Where pushback mining is being carried out the ore loss is adjusted to allow for ore spill over on the high wall. The ore spill over loss adjustment is an interim adjustment and is reviewed when the pit floor level reaches the ore spill material. Although there are short term adjustments in dilution and ore loss, historical results support these values.

18^th September 2018 Page 210

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Underground Operations

      Loulo

The Loulo site technical team routinely reconcile actual stope and development production against planned performance. A summary of the results is shown in Table15-12 and Table15-13.

Two forms of dilution have been included in the 2017 Ore Reserve estimate, planned dilution (inside the stope design solid) and unplanned dilution (outside the stope design solid).

Planned dilution is included where the footwall or hanging wall is folded and dilution is added to create a mineable stope shape. In Yalea, this is a small quantity (adding 3% dilution); however, in Gara, it is more significant (adding 5% dilution) due to some narrow and highly folded areas. Dilution is added at a gold grade of 0.00 g/t.

Unplanned dilution is added as an equivalent thickness of waste on the hanging wall as panel widths vary in Yalea (3 m to 28 m, average 12 m) and Gara (3 m to 23 m, average 7 m). The Yalea fresh rock stope walls are quite competent with a clean break. The Gara stope walls are less competent, but blasting is carefully managed by the on site teams. This contributes to a relatively low level of dilution

Unplanned dilution thicknesses were calculated from reconciliation of stope performance. All stopes mined in 2017 had a fresh rock hanging wall so this result was used for the fresh rock hanging wall dilution parameter, after subtraction of an allowance for improvement in mining practices.

Paste dilution is now being tracked separately. In 2017, Yalea had 4% paste dilution and Gara 2%. As a result of this understanding of the significance of paste dilution, the paste fill strengths in the lower portion of the wide stopes are being reviewed by the site technical team.

Overall, waste actual dilution was estimated to be an equivalent hanging wall thickness of 0.82 m in Yalea and 1.58 m in Gara. These include planned dilution (within the stope shape) and unplanned dilution (outside of the stope shape).

Table15-12 Summary of Actual 2017 Loulo Underground Mining Dilution

Operation	Dilution Reported as %s					Dilution Reported as Thickness			Total Dilution
	Overbreak Dilution			Fill Dilution (CAF & Paste)		Actual
	Total	Un- Planned	Planned	CAF Dilution	Paste Dilution	Total	Un- Planned	Planned
	%	%	%	%	%	m	m	m	%
Yalea	11	7	6	0	2	0.82	0.45	0.38	13
Gara	20	20	5	1	4	1.58	1.26	0.31	25

18^th September 2018 Page 211

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-13 Summary of Loulo Underground Ore Reserve Estimate Dilution

Area	Mining Method	Rock Type	Planned Dilution (%)	Unplanned Dilution (%)	Unplanned Dilution Thickness (m)	Total Dilution (%)	Total Dilution Thickness (m)	Average Stope Thickness (m)
Yalea	Long Hole Stoping	Fresh Ore & Hanging Wall	2.7	9.1	1.2	12	1.6	11.7
Yalea - South Upper	Stoping Under Rock Fill	Fresh Ore	1.9	35.0	4.0	37	4.3	11.3
Yalea - South Upper	Stoping Under     Rock Fill	Partially Weathered Trans Ore	3.6	35.0	5.0	39	5.5	13.8
Yalea - South Upper	Stoping Under Rock Fill	Fully Weathered Oxide Ore	3.6	35.0	5.0	39	5.4	13.8
Yalea Average			3.0	9.4	1.7	16	2.0	11.8
Gara	Long Hole Stoping	Fresh Ore &     Hanging Wall	5.0	10.9	1.0	16	1.5	7.7

Ore loss in the 2017 Ore Reserve estimate has been included as 4% for Yalea stopes with fresh rock hanging wall and 4% for Gara stopes with fresh rock hanging wall. Yalea ore losses were increased from 2% in the previous estimate to reflect the increase observed in 2017 relative to 2016. Overall actual ore loss is currently estimated to be 5% in Yalea (8% in 2017, 5% in 2016). At Gara actual ore loss is currently 4.5% (2% in 2017, 8% in 2016).

Part of the upper south area of the Yalea has weathered “transition” and oxide material in the stopes and hanging wall. Where transition or oxidised rock was present in the stope or hanging wall rock within 5 m of the stope additional ore loss has been included. This area is planned to be mined using a stoping under rock fill method. Overall ore loss in the stoping under rock fill mining area is 32%. The actual dilution and ore loss values measured and the parameters used for the Ore Reserve estimate are summarised in Table15-14 and Table15-15.

Table15-14 Summary of Loulo Underground Ore Reserve Estimate Dilution

		Estimated Av. Vein	Ore
UG	Mining	Thickness	Loss
Deposit	Method	(m)	(%)
	LHOS	12.7	4.0
	LHOS	10.6	4.0
Yalea	SURF	13.0	32.3
	LHOS+ SURF	11.0	8.9
		11.5	7.6
	LHOS	7.6	4.0
Gara	LHOS	7.8	4.0
	LHOS	7.7	4.0

Table15-15 Summary of Loulo Underground Ore Reserve Estimate Dilution

		Estimated Av. Vein	Ore
UG	Mining	Thickness	Loss
Deposit	Method	(m)	(%)
	LHOS	12.7	4.0
Yalea	LHOS	10.6	4.0
	SURF	13.0	32.3

18^th September 2018 Page 212

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

	LHOS+         SURF	11.0	8.9
		11.5	7.6
	LHOS	7.6	4.0
Gara	LHOS	7.8	4.0
	LHOS	7.7	4.0

Gounkoto Dilution and Ore Loss Assessment

The mining method selected for Gounkoto underground is longhole stoping and longhole with backfill and incorporates both transverse and longitudinal access as appropriate to the width of the orebody. Ground conditions have been estimated to be in the majority ‘Fair’, with minor inclusions of ‘Poor’ ground. Stope designs have been based on a stope stability assessment for stable stope surfaces. Therefore, unplanned dilution is expected to be low provided good mining practice is followed.

Stopes have been designed for longhole stoping using a “top hammer” blast hole drill. The complexity of the orebody required some waste, Inferred and unclassified material be included in stope designs so that suitable recovery of the mineralisation abovecut-off grade is achieved. Inferred and unclassified material is classed as planned dilution. None of the proposed stopes report below 3.0 g/t Au and the proportion of Inferred and unclassified material that is included in stope designs is below 1.0% of the total stope tonnes which is considered to be acceptable.

Dilution and losses have been modified where blind uphole stopes are used because of the reduced flexibility in blast hole drilling geometry. A summary of the expected dilution and losses is given in Table15-16.

Table15-16 Gounkoto Underground Mining Dilution and Losses

Mining Method	Dilution (%)	Losses (%)
Transverse Primary Stopes	5	5
Transvers Primary Stopes (blind uphole)	7.5	8
Transverse Secondary Stopes	7.5	5
Transverse Secondary Stopes (blind uphole)	10	8
Longitudinal Stopes	10	5
Longitudinal Stopes (blind uphole)	10	21

Dilution for the blind uphole stopes is higher than usual stopes to take account of the more difficult mining conditions. In the transverse secondary stopes, waste dilution can occur from the hanging wall contact and the cemented fill, which makes this dilution higher than in the primary stopes.

Similar adjustments were made to the expected mining losses. Expected dilution in the longitudinal stopes is 10% and mining losses are 5%.

Recovery for blind uphole longitudinal stopes has been set at 79% which accounts for 5 m rib pillars being left every 25 m and 95% recovery of the remaining stopes. Recovery for longitudinal blind uphole stopes is better than transverse blind uphole stopes because of better blast hole drilling geometry in narrower stopes.

18^th September 2018 Page 213

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

15.10 Economic Parameters

Differentcut-off grades were applied to the operations and projects as each is subject to different operating costs, losses, and recoveries. Thecut-off grades discussed here were based on simple operating cost estimates, used in the design work, and subsequently confirmed to be viable by a cash flow calculation based on the production schedule and full set of costs.

Open Pit Criteria

The calculation of thecut-off grade is based on the premise that mining has already exposed the ore and the only decision that remains is whether to send the material to the crusher or the waste dump. Mining is viewed as a sunk cost and the cost of exposing the ore is not incorporated into thecut-off grade calculation for material within the pit. While the calculation of the mining cost used in the optimisation process was adjusted for depth based on the 2016 pit designs; the processing cost was not adjusted for depth or variation between ore and waste. The adjustments for additional fuel usage when hauling from depth, additional pumping cost as depth increases and the extra costs associated with drilling and blasting ore are included in the mining cost adjustment factors in the optimisation process.

The full gradecut-off is the ore material that is profitable at $1,000/oz, considering all operating costs of mining, haulage, processing, and general and administrative costs, as well as the appropriate recovery, dilution, and realised gold price post royalty at $1,000/oz spot gold price. It is the principal material fed to the plant.

The marginalcut-off grade is the ore material mined within the $1,000/oz pit design that is profitable on a marginal basis (Full operating cost minus mining costs). The plan is to blend this material with the high-grade ore from the underground mine. Both marginal and full grade ore form the reserve since it forms part of the LOM plan feed schedule.

The marginal orecut-off grade is the reservecut-off grade.

Loulo Open Pits

The weighted averagecut-off grades for the Gara West, Baboto and Loulo 3 open pits are 1.09 g/t Au, 0.96 g/t Au, and 1.09 g/t Au respectively.

Table15-17 shows the economic parameters used for calculating thecut-off grades for the Loulo open pits to determine the Ore Reserves.

Mining costs are based on historical costs for similar size operations within Randgold. The Baboto mining cost is based on tender costs received from SFTP, a Malian mining contractor.

Table15-17: Loulo Open Pit Economic Parameters

Parameter	Unit	Gara West	Baboto	Loulo 3
Gold Price	$/oz		1,000
Royalty	%		6%
Selling cost	%		0%

18^th September 2018 Page 214

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Parameter	Unit	Gara West	Baboto	Loulo 3
Net Gold Price	$/oz	940
Met Recovery	%	91%
Dilution	%	10%
Ore Loss	%	2%
Mining Cost - Contractor	$/t mined	3.63	3.18	3.45
Mining Cost - Owner’s	$/t mined	0.06	0.06	0.06
Mining Cost - Grade Control	$/t mined	0.21	0.07	0.21
Total Mining Cost	$/t mined	3.90	3.31	3.72
Strip Ratio	Waste/Ore	1.94	4.5	1.94
G&A	$/t milled	8.80	7.80	8.80
Ore Crushing & Hauling	$/t milled	0	3.63	0
Mining	$/t milled	11.46	18.18	10.93
Process Plant	$/t milled	20.6	15.2	20.6
Total Operating Costs	$	40.86	44.81	40.33
Full GradeCut-off	g/t	1.52	1.62	1.50
MarginalCut-off Grade	g/t	1.09	0.96	1.09
Diluted Full GradeCut-off Grades	g/t	1.38	1.48	1.36
Diluted MarginalCut-off Grades	g/t	0.99	0.88	0.99

18^th September 2018 Page 215

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto Open Pit

In the open pit, the weighted averagecut-off grades for the Gounkoto and Faraba open pits are 1.20 g/t Au and 1.32 g/t Au respectively. Table15-18 shows the economic parameters used for estimating thecut-off grade for the Gounkoto open pit and Faraba open pit, in determining the Ore Reserves.

Table15-18: Gounkoto Open Pit Economic Parameters

Parameter	Unit	Gounkoto	Faraba
Gold Price	$/oz	1,000
Royalty	%	6%
Selling cost	%	0%
Net Gold Price	$/oz	940
Met Recovery	%	91%
Dilution	%	10%
Ore Loss	%	2%
Mining Cost - Contractor	$/t mined	2.95	3.45
Mining Cost - Owner’s	$/t mined	0.06	0.06
Mining Cost - Grade Control	$/t mined	0.07	0.14
Total Mining Cost	$/t mined	3.08	3.65
Strip Ratio	Waste/Ore	13.62	3.61
G&A	$/t milled	7.70	8.80
Ore Crushing & Hauling	$/t milled	5.89	5.9
Mining	$/t milled	44.98	16.81
Process Plant	$/t milled	19.0	20.6
Total Operating Costs	$	77.57	52.11
Full GradeCut-off	g/t	2.85	1.95
MarginalCut-off Grade	g/t	1.20	1.32
DilutedCut-off Grades	g/t	2.59	1.77
Diluted MarginalCut-Off Grades	g/t	1.09	1.20

Underground Criteria

The calculation accuracy of thecut-off grade for the Loulo underground deposits was different to that of the Gounkoto deposit, as the Gounkoto work is based on lower confidencepre-feasibility level data.

Loulo Underground

For the Loulo deposits, a full cost minedcut-off grade was estimated, including all operating costs and capital requirements. Revenue was factored down by the processing recovery. Mine dilution and mining recovery were not included in this calculation as thiscut-off grade was applied to mined material to provide an indication of the average grade of a mine, or mine area, required to cover all costs. It was not used to exclude material from the Ore Reserve.

A Marginal minedcut-off grade was estimated in a similar way to the full costcut-off grade, however, it included only the extra operating costs that would be incurred if a stope was mined, and so excluded fixed mining and capital costs. Thiscut-off grade provided an indication if a decision to include (or exclude) a panel in the Ore Reserve would add value. This is the Ore Reservecut-off grade.

The full cost of processing and the site administration cost have been included in the calculation of the marginalcut-off grade. This was included because the Loulo process plant is expected to

18^th September 2018 Page 216

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

be fed to its capacity, and there are multiple ore sources available to provide feed to the Loulo process plant (Yalea, Gara, Gounkoto, and satellite deposits). Any additional lower grade material (just below marginalcut-off grade) would likely displace higher grade mill feed.

In addition, insitucut-off grades were calculated for use when evaluating or viewing block model grades. These were the same as the minedcut-off grades but included mine dilution and recovery.

Cost analysis was undertaken on 2017 actual total mining costs to provide a basis forcut-off grade and economic analysis for the Ore Reserves estimate.

Marginal and break-evencut-off grades were calculated for both Yalea and Gara. Thecut-off grade used for the estimation of the December 2017 Ore Reserves is the diluted marginalcut-off grade (2.60 g/t Au for Yalea and 2.30 g/t Au for Gara). Thesecut-off grades were applied to stope panels after dilution and ore loss had been accounted for in the panel. In Yalea a 2.60 g/t Au cut-off grade is used for the stoping under rock fill (SURF) mining area due to lower backfill costs.

The inputs to thecut-off grade calculations are shown in Table15-19.

18^th September 2018 Page 217

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-19 Yalea and Gara Underground Mine –Cut-Off Grade Calculation

Parameter	Unit	Yalea	Gara
Gold Price	$/oz	1,000	1,000
Royalty	%	6%	6%
Selling cost	%	0%	0%
Net Gold Price	$/oz	940	940
Met Recovery	%	84.47%	94.2%
Dilution	%	11.5%	15.6%
Ore Loss	%	2%	4%
Mine OPEX Development	$/t milled	5.00	3.29
Mine Stoping	$/t milled	12.15	11.79
Backfill	$/t milled	11.58	10.51
Fixed Cost	$/t milled	13.98	14.63
Grade Control	$/t milled	3.37	3.40
Mine Sustaining Capital	$/t mined	7.69	8.25
Total Mining Cost	$/t mined	53.78	53.78
G&A	$/t milled	7.80	7.70
Process Plant	$/t milled	17.80	18.20
Total Operating Costs	$	79.38	79.68
BreakevenIn-SituCut-off Grade	g/t	3.54	3.29
Breakeven MinedCut-off Grade	g/t	3.11	2.73
MarginalIn-situCut-off Grade	g/t	2.97	2.80
Marginal MinedCut-off Grade	g/t	2.61	2.33

Gounkoto Underground

Underground mine operating costs were determined using November 2012 contractor mining costs, which formed a basis for the December 2016cut-off grade for the Ore Reserve estimate. An insitucut-off grade of 3.0 g/t Au was used for the stope design in the main part of the orebody, and an insitucut-off grade of 4.0 g/t Au was used for the stope design in more isolated areas of the orebody. No supporting calculation was provided for thesecut-off grades. As a check, a new and currentcut-off grade was determined, resulting in a marginal insitucut-off grade of 3.34 g/t Au. This is approximately in line with the combined 3.0 g/t Au and 4.0 g/t Aucut-off grade and compares favourably to the marginal insitucut-off grades of Yalea and Gara of 2.97 g/t Au and 2.80 g/t Au, which are similar operations. RPA recommends the Gounkoto mine design is adjusted with the newcut-off grade results.

The Gounkoto underground ore will be toll treated at the Loulo mill. The total processing cost applied is based on the current costs for the Loulo mill. The inputs to the revisedcut-off grade calculations are shown in Table15-20.

18^th September 2018 Page 218

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-20 Gounkoto Underground Mine –Cut-Off Grade Calculation

Parameter	Unit	Gounkoto
Gold Price	$/oz	1,000
Royalty	%	6%
Selling cost	%	0%
Net Gold Price	$/oz	940
Met Recovery	%	92.0%
Dilution	%	710.2
Ore Loss	%	6.3%
Mine Capital Development	$/t mined	34.36
Total Mining Cost	$/t mined	49.52
G&A	$/t milled	9.80
Process Plant	$/t milled	23.60
Total Operating Costs	$	117.28
BreakevenIn-SituCut-off Grade	g/t	4.96
Breakeven MinedCut-off Grade	g/t	4.22
MarginalIn-situCut-off Grade	g/t	3.34
Marginal MinedCut-off Grade	g/t	2.84

15.11 Pit Optimisations

Open pit planning was completed using Geovia Whittle 4.1 to derive the optimal pit shell and Geovia Surpac 6.7 for the detailed engineering and design work on the optimised pit shells.

Economic parameters and physical constraints were input into the optimisation software to generate a series of nested pits from which an optimal shell was selected. The reserve pit optimisation was completed in two stages. The initial pit optimisation excluding a minimum mining width was completed to select the optimum pit shell. Thereafter, an evaluation with a minimum mining width of 75 m was run to define the pushbacks for pit scheduling.

Loulo Open Pits

The Loulo 3 pit optimisation was updated for the 2016 Reserve using the “Loul3_res_2015_11_12” resource model combined with the 2016 mining and processing costs and processing recoveries. No updates were undertaken on the 2016 resource model with no significant changes observed on mining and process costs or other economic optimisation parameters. The 2016 Loulo 3 reserve was therefore carried through to the 2017 Reserve.

The Pit Optimisation pit by pit graph in Figure15-2 shows a very flat cash flow generation between the $900 (Pit 5) and the $1,100 (Pit 7) pits, indicating no possible gain in selecting a higher gold price pit than the $1,000 reserve pit. This also indicates there is a fairly low risk associated with a 10% swing in gold price.

18^th September 2018 Page 219

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-2 Pit by Pit Graph of Whittle Optimisation Results for Loulo 3

The results shown in Table15-21 below give further validation to the decision to remain with a $1,000 Whittle shell as the basis for the 2016/2017 Ore Reserve estimate.

Table15-21 Whittle Optimisation Results for Loulo 3

Pit Gold Price ($/oz)	Cash Flow ($ M)	Ore Tonnes (Mt)	Waste Tonnes (Mt)	Ounces (koz)	Grade (g/t)
500	21.46	0.24	0.81	36.91	4.80
600	36.98	0.50	3.56	75.73	4.71
700	43.00	0.65	5.34	95.57	4.60
800	81.02	1.38	24.09	245.85	5.53
900	89.66	1.98	34.37	322.63	5.08
1,000	91.70	2.23	40.02	358.88	5.00
1,100	91.14	2.38	42.48	374.74	4.90
1,200	89.21	2.49	45.14	388.83	4.85
1,300	84.05	2.69	49.69	410.71	4.75
1,400	47.54	3.71	76.69	533.13	4.47
1,500	40.20	3.85	81.20	551.03	4.45

The Baboto pit optimisation was updated for the 2017 Reserve using the “BS_BC_Regularize Model_170514” resource model. The most notable change from the 2017 resource model to the 2016 model was the exclusion of the Baboto North zone which was acquired by Endeavour Mining in 2017.

The pit optimisation pit by pit graph in Figure15-3 and the results in Table15-22, show a relatively flat cash flow generation between the $900 (Pit 5) and the $1,200 (Pit 8) pits, indicating no significant gain in cash flow or ounces if a higher gold price pit is selected. The Baboto optimisation also indicates there is a fairly low risk associated with a 10% swing in gold price.

18^th September 2018 Page 220

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-3 Pit by Pit Graph of Whittle Optimisation Results for Baboto

Table15-22 Whittle Optimisation Results for Baboto

Pit Gold Price ($/oz)	Cash Flow ($ M)	Ore Tonnes (Mt)	Waste Tonnes (Mt)	Ounces (koz)	Grade (g/t)
500	24.50	0.54	0.69	51.70	3.00
600	34.40	0.85	1.44	78.40	2.86
700	46.60	1.37	3.31	119.90	2.73
800	54.40	1.75	5.94	154.70	2.74
900	56.70	2.00	7.17	172.00	2.67
1,000	57.30	2.20	8.02	183.70	2.60
1,100	56.80	2.39	9.53	196.60	2.56
1,200	54.90	2.59	11.23	209.10	2.51
1,300	53.00	2.73	12.52	217.60	2.48
1,400	50.80	2.83	13.64	223.60	2.46
1,500	47.70	2.94	15.04	230.30	2.44

The Gara West optimisation was completed in 2015 based on the “gw_res_2015_11_24_bm.bmf” resource. No further updates have been carried out on the resource and optimisation. The pit optimisation pit by pit graph in Figure15-4 and results in Table15-23 indicate a robust $1,000 pit that will allow for a 10% fluctuation in gold price.

18^th September 2018 Page 221

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-4 Whittle Optimisation Results for Gara West

Table15-23 Pit by Pit Graph of Whittle Optimisation Results for Gara West

Pit Gold Price ($/oz)	Cash Flow ($ M)	Ore Tonnes (Mt)	Waste Tonnes (Mt)	Ounces (koz)	Grade (g/t)
500	12.06	0.18	0.26	22.65	3.91
600	13.95	0.24	0.36	27.73	3.58
700	18.03	0.40	0.87	4.39	3.19
800	20.24	0.54	1.26	51.39	2.94
900	21.96	0.72	2.76	67.77	2.90
1,000	22.69	0.95	3.87	82.95	2.71
1,100	22.33	1.06	4.75	90.98	2.67
1,200	19.20	1.36	7.64	113.52	2.59
1,300	11.37	1.82	12.51	147.02	2.51
1,400	-0.73	2.33	18.22	182.41	2.43
1,500	-10.85	2.68	22.21	205.38	2.38

Gounkoto Open Pits

The Gounkoto 2017 pit optimisation was completed using the “gk_gc_170714.bmf” resource model. The optimisation, when compared to the previous 2016 optimisation, showed a gain in ore tonnes due to the inclusion of the hanging wall extension ore within the $1,000 Whittle pit.

Figure15-5 shows the area with the extra ore tonnes relative to the current life of mine and reserve pit design.

18^th September 2018 Page 222

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-5 Gounkoto Pit Design Oblique View looking upwards from underneath, with Gain in Ore Tonnes

This hanging wall extension contains 1.1 Mt of ore, of which 0.85 Mt is below the full grade cut-off, so most of this material would be stockpiled with no short to medium term gain in plant feed and revenue. As such, this material was excluded from the 2017 Ore Reserve statement and further drilling is being carried out in 2018.

The optimisation pit by pit graph in Figure15-6 and Table15-24, show that the Gounkoto Super Pit is still robust at $1,000, allowing for 20% fluctuations in the gold price without significant impact on the economic viability of the operation.

Figure15-6 Whittle Optimisation Results for Gounkoto

18^th September 2018 Page 223

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-24 Whittle Optimisation results for Gounkoto

Pit Gold Price ($/oz)	Cash Flow ($ M)	Ore Tonnes (Mt)	Waste Tonnes (Mt)	Ounces (koz)	Grade (g/t)
500	551.90	4.69	27.60	910.50	6.04
600	616.50	6.04	37.65	1077.00	5.54
700	889.50	11.62	150.01	2013.90	5.39
800	923.30	12.78	169.41	2169.40	5.28
900	939.70	13.81	190.81	2309.70	5.20
1,000	945.40	15.09	204.08	2412.60	4.97
1,100	943.40	15.62	213.09	2465.10	4.91
1,200	931.80	16.41	229.58	2543.10	4.28
1,300	920.20	16.88	242.21	2595.30	4.78
1,400	887.90	18.01	266.13	2693.90	4.65
1,500	875.90	18.33	274.13	2721.40	4.62

The Faraba optimisation pit by pit graph shows that the $1,000 pit (pit 6) is stable at $1,000 and can withstand a 10% variation in gold price, however, the optimisation does indicate that should the gold price drop to $800/oz there is virtually no economic pit.

Figure15-7 Whittle Optimisation Results for Faraba

Table15-25 Whittle Optimisation Results for Faraba

Pit Gold Price ($/oz)	Cash Flow ($ M)	Ore Tonnes (Mt)	Waste Tonnes (Mt)	Ounces (koz)	Grade (g/t)
500	0.28	0.01	0.00	0.58	3.02
600	1.89	0.05	0.04	4.50	2.68
700	2.61	0.08	0.07	6.75	2.52
800	8.44	0.52	1.45	38.62	2.33
900	31.84	2.59	7.63	187.56	2.26
1,000	32.31	2.74	8.05	197.41	2.24
1,100	31.11	3.27	10.51	232.25	2.21
1,200	30.13	3.39	11.20	239.60	2.20
1,300	28.02	3.51	11.86	246.08	2.18
1,400	24.80	3.67	13.12	255.43	2.17
1,500	22.61	3.73	13.81	259.35	2.16

18^th September 2018 Page 224

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

15.12 Mine Design

Open Pits

Detailed design was then undertaken to confirm the viability of the optimised shell. The relatively small dimension of the pits and the characteristics of the ore bodies (i.e. narrow and steeply dipping Birimian meta-sedimentary volcanic sequence), combined with the equipment size is used to engineer the pit design to keep strip ratios in line with optimisation results. The process was iterated until an acceptable level of correlation was achieved between the optimised shell and detailed design.

Mine planning was based on three-dimensional block models of insitu mineralisation, with allowances made for minimum mining widths, dilution, and ore loss appropriate to the mining method being considered. Historical performance measures were considered in determination of these modifying factors.

Infrastructure, waste disposal, and ore stockpile management requirements were incorporated into the planning process. In general, the pit designs can be typified by the characteristics found in Table15-26.

Table15-26 Open Pit Design Characteristics

Item	All Open Pits
Accesses	Minimum of two accesses at all times
Ramps	Two single lane ramps (1 driving down, 1 driving up)
Ramp Access	Minimising ramp access in the hanging wall, unless in a temporary wall
Ramp Width	25 m for double lane and 12 m for single lane ramps, except for dumps (30 m)
Ramp Gradient	10%
Batter Angles	60-75°
Switchbacks	Kept to a minimum
Geotechnical Recommendations	Followed
Bench Basis	10 m bench (except for Loulo 3 in the saprolite, with 5 m benches planned)
Strip Ratio	Kept as low as possible
Economic Testing	Tested with Whittle 4X
Final Face Benches	No ramp access
Single Lane Ramps	Used where required

The Baboto resource model was updated during 2017. The updated model was used to update the optimisation and pit design along with a mining schedule. A sketch of the updated mine design is given in Figure15-8.

18^th September 2018 Page 225

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-8 Baboto Pit and Waste Dump Design

The Gara West pit is a simple pit with similar geotechnical characteristics to Gara, which is nearby. The Whittle optimisation and design accounts for the impact of the western waste dump on the pit size (Figure15-9).

18^th September 2018 Page 226

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-9 Gara West Pit And Waste Dump Design

Loulo 3

The Loulo 3 pit has been designed to allow for a pushback to reach deeper ore based on the $1,000/oz Whittle optimised pit (Figure15-10). Utilising smaller equipment sizes in the mining fleet has allowed for narrower ramps and a narrower mining width thereby reducing the amount of waste to be mined and the overall mining cost.

Figure15-10 Loulo 3 Pit Design

18^th September 2018 Page 227

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Underground Mine Design

Loulo

The 2017 Loulo underground mine plan design and evaluation was completed using Datamine 5D planner software using sub cell block models. The geological zones (including mineralised zone) were defined by three-dimensional wireframe solids and surfaces. Both the block models and wireframes were created with Maptek Vulcan software and converted into Datamine format for use in Datamine 5D planner.

The following process was followed to estimate the Ore Reserve:

1. Cut-off grades were determined from actual 2017 costs and the previous life of mine plan.

2. Stope section strings (2.5 m, 5 m, or 10 m interval between sections) were created to follow the geological mineralised zone wireframes. Strings are based on level intervals determined in the previous Ore Reserve estimate.

3. Planned dilution was included where the footwall or hanging wall was folded and dilution is added to create a mineable stope shape.

4. Stope section strings were edited based on the as built mined solids to remove development drives and parts of stopes.

5. Stope wireframes were created from the strings using Datamine 5DP software.

6. The total stope tonnes and gold metal was calculated by evaluating the stope wireframes against the block model.

7. Diluted mined tonnes, grades, and ounces were calculated in Datamine EPS scheduler. This included unplanned dilution added as an equivalent thickness on the hanging wall (or as a percentage of waste for the SURF method). Ore Loss is subtracted as a percentage from diluted tonnes and contained metal.

8. Final editing of physical quantities was undertaken in an Excel spreadsheet.

9. Panels with a diluted grade belowcut-off were excluded from the Ore Reserve estimate. A level by level cash flow analysis was used to confirm that each level contributed positive cash flow after decline and access development was taken into account. Levels with a negative cash flow were excluded from the Ore Reserve.

10. Classification of the Ore Reserve was reviewed and, where necessary, adjusted to ensure that classification of Ore Reserves reflects the QP’s view of the deposits.

The location of the Loulo underground Ore Reserves is shown in Figure15-11 and Figure15-12.

18^th September 2018 Page 228

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-11 Yalea Mine Design with Ore Classification

Figure15-12 Gara Mine Design with Ore Classification

Gounkoto

The 2017 Gounkoto underground Ore Reserve estimate was completed using Datamine Studio 5D Planner software with the Mineable Stope Optimiser (MSO) module. This process provided a

18^th September 2018 Page 229

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

comprehensive understanding of the location of the potential stoping areas and their respective tonnes and grade.

An MSO run was conducted on the resource model below the northern part of the open pit. Some of this material was in the Indicated category.

The principle MSO parameters are shown in Table15-27. Long hole stoping method parameters were used. The minimum mining width was 3.0 m and the minimum dip for stopes was set to 50°. The maximum width for a stope was set to 100 m to represent an unlimited maximum mining width, as in practice these stopes would be split into a hanging wall and footwall stopes and mined and filled sequentially where the horizontal width is beyond stable geotechnical widths. The minimum width of internal pillars of material that are below thecut-off grade was set to 10 m. This pillar width assumption was reviewed and confirmed for the final stope designs.

A head gradecut-off of 3.0 g/t Au was used in the MSO work, based on the November 2012 Underground Mining Study. Thiscut-off grade is further detailed in Section 15.8.

A longitudinal projection of the results is shown in Table15-27.

Table15-27 MSO Parameters for MSO Run

Item	MSO Setting	Comment
Shape Control	Cut-Off Grade (g/t)	3.0
	Width (m)	3
	Max Width (m)	100
	Min Waste Pillar Width (m)	10
	Min Dip Angle (°)*	50
	Max Dip Angle (°)*	130
	Max Strike Angle (°)	45
	Max Strike Angle Change	20
	Max Side Length Rati	2.25
Shape Waste Control	Maximum Waste Faction	1
Model Discretisation	U Number	4
	V Number	4

18^th September 2018 Page 230

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure15-13 Longitudinal Projection Looking West showing Results of MSO Run on Northern End of the Resource Model

15.13 Ore Reserve Comparisons

JORC compliant Reserve statements were issued in 2016. Various changes were made to these Reserves up to the end of 2017, as detailed in this section. Section 6.4 (Resource and Reserve Evolution) illustrates changes to the Reserves over the last few years

Loulo Open Pit

The overall 2017 declared ounces for Loulo is 14% lower than 2016. This is predominantly due to the sale of the Baboto North resource, accounting for 76 koz with the remainder due to model and density changes in the Baboto pit. No mining was carried out in the other pits and stockpiles (Table15-28).

18^th September 2018 Page 231

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table15-28 Loulo Open Pits Ore Reserve Reconciliation

Change	Gara West     (koz)	Baboto     (koz)	Loulo 3     (koz)	Stockpiles     (koz)	Total Loulo Open Pits     (koz)
2016 Declaration	73	291	348	93	805
2017 Depletion
Model Change		-9.1
Density change		-23
Pit Design change		-1.8
Sale of Baboto North		-76
Economic change		3.8
Adjustment		-5.4
2017 Declaration	73	180	348	93	694

Gounkoto Open Pit

The 2017 declared ounces are 7% lower than the 2016 declared ounces for the Gounkoto open pit. This is due to depletion in 2017, with a partial replenishment from model gains due to continued grade control drilling.

There is no change reported for the Faraba open pit (Table15-29).

Table15-29: Gounkoto Open Pit Ore Reserve Sources Reconciliation

Change	Gounkoto Open Pit     (koz)	Faraba Open Pit     (koz)
2016 Declaration	2,509	200
2017 Depletion	-324
Model Change	155
Density Change
Economic Change
Design Change	6
2017 Declaration	2,334	200

Loulo Underground

During 2017, a significant amount of resource extension and grade control drilling was completed in both Yalea and Gara. This led to addition of new areas in both Yalea Far South and Gara Far South Extension (450 koz and 500 koz respectively). Grade control drilling from underground drill sites led to reinterpretation of existing Ore Reserves and a loss in previous Ore Reserves areas of 190 koz in Yalea and 30 koz in Gara.

The net effect was increase in the Yalea Ore Reserve of 180 koz (+6%) and an increase in the Gara Ore Reserve by 390 koz (+21%), after depletion was taken into account.

15.14 External Audits

Loulo

During 2013 and 2014 the Yalea and Gara Life of Mine Planning was reviewed by Practical Mining (Australia). The review concluded that the current mine planning was robust but

18^th September 2018 Page 232

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

recommended further work to improve confidence in mine ventilation, paste fill testwork, geotechnical rock mass modelling,pre-mining stress measurement and mining induced stress modelling. Randgold has undertaken substantial work in these areas and this work is ongoing.

18^th September 2018 Page 233

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

16  Mining Methods

16.1  Loulo Open Pits

The Loulo open pits comprise the Baboto Pit, Loulo 3 Pit, and the Gara West Pit. Some of the Baboto pit was mined in 2018, however the rest of Baboto pit and the other pits will be mined from 2024, after the larger Gounkoto pits start to deplete. Due to the late start date, not much of the planned production has been detailed yet. Randgold intends to use the same contractor that has already mined part of the Baboto pit, so most of this fleet is known. The Loulo 3 and Gara West pits will use the same equipment as they are timed to be mined consecutively.

Baboto

The strategy for the mining of Baboto is to mine a small oxide material pit in 2018 in a first phase and then to return and mine the remainder of the full reserve pit in 2027 in a second phase. A total 279 kt grading 3.12 g/t Au was mined from the oxide pit in 2018 and hauled to the Loulo ROM pad. The rest of the pit will be mined to the final pit limit in 2027, for a total of 2.0 Mt at 2.82 g/t Au and is feed for the Loulo plant from 2027 to 2029 (Table16-2).

Mining is planned to be by mining contractor. Table16-1 lists the planned mining equipment for the Baboto open pit.

Table16-1 Baboto Main Mining Equipment

Mining Equipment	2018	2027	Equipment Type
8 m3Hydraulic backhoe excavator	2	2	Liebherr 970
40 T Articulated Dump Trucks	5	8	Volvo A45G
Pre-split drill rig - 115 mm diameter	0	2	To be decided
Production drill rig –165 mm diameter	0	2	To be decided
Grade control drilling	0	1	Schramm T450B
Loader CAT 980 or equivalent	1	1	Caterpillar 980K
33 m3Haulage truck	5	5	Volvo FMX 440

Figure16-1 shows the location of the Baboto open pits, ROM pad, waste dumps. The Baboto pits are located 12.2 km from the Loulo plant. A haul road has been constructed to haul the ore to the Loulo plant for processing (Figure16-2).

18^th September 2018 Page 234

18^th September 2018 Page 235

18^th September 2018 Page 236

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Life of Mine Plan

The mining of the Loulo satellite pits is planned to commence after the completion of the Gounkoto open pit. The Loulo 3 open pit mining will be a pushback from the existing open pit which completed mining in 2011. The additional pushback has become economic on the back of additional resource drilling which detected additional high-grade material in the MZ2 ore zone, below the southern end of the pit.

The Baboto oxide ore will be mined in 2018, to supplement the Loulo plant throughput and will then stop until 2027 when the remaining sulphide ore will be mined.

The Gara West Pit is situated adjacent to the mined out Gara open pit and is in close proximity to the existing infrastructure.

Following on Randgold’s initiatives to promote local business growth, it is envisaged that local mining contractors will be used to mine the Loulo satellite pits.

Table16-2 details the Loulo mine production plan across all open pit ore sources.

Table16-2 Loulo Open Pit Life of Mine Plan

Item	Units	LOM	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027
Open Pit Mining
Waste
Loulo 3	kt	45,901	-	-	-	-	-	-	19,163	26,738	-	-
Baboto	kt	9,485	582	-	-	-	-	-	-	-	-	8,904
Gara West	kt	7,396	-	-	-	-	-	-	-	-	7,396	-
Total Waste Tonnes	kt	62,782	582	-	-	-	-	-	19,163	26,738	7,396	8,904
Ore
Loulo 3	kt	2,202	-	-	-	-	-	-	977	1,225	-	-
Baboto	kt	2,311	337	-	-	-	-	-	-	-	-	1,974
Gara West	kt	862	-	-	-	-	-	-	-	-	862	-
Total Ore Tonnes	kt	5,375	337	-	-	-	-	-	977	1,225	862	1,974
Gold Grade
Loulo 3	g/t	4.99	-	-	-	-	-	-	3.75	5.97	-	-
Baboto	g/t	2.79	2.64	-	-	-	-	-	-	-	-	2.82
Gara West	g/t	2.62	-	-	-	-	-	-	-	-	2.62	-
Total	g/t	3.66	2.64	-	-	-	-	-	3.75	5.97	2.62	2.82
Gold Ounces
Loulo 3	koz	353	-	-	-	-	-	-	118	235	-	-
Baboto	koz	208	29	-	-	-	-	-	-	-	-	179
Gara West	koz	73	-	-	-	-	-	-	-	-	73	-
Total	koz	633	29	-	-	-	-	-	118	235	73	179
Strip Ratio	w:o	11.7	1.7	-	-	-	-	-	19.6	21.8	8.6	4.5

16.2 Gounkoto Open Pit Operations

Mining

Gounkoto is an ongoing operation with a successful selective mining method. Backhoe excavators are used to select waste from the ore. Pit geologists supervise all digging and ore material is classified by grade as either FGO (Full Grade Ore), delivered to primary crushers, or MO (Marginal Ore), which is stockpiled close to the primary crushers. Waste material is hauled

18^th September 2018 Page 237

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

to the nearest waste dump. Truck allocation is completed automatically through the use of the Jigsaw dispatch system. Management reporting is via an intranet-based system.

Fresh rock and transitional zones are drilled and blasted in 10 m lifts, with excavation of benches containing ore in three flitches. A small volume of weathered oxide material is free dig.

Blasting is based on a 5 m by 5 m grid with a drill-hole diameter of 165 mm and a powder factor of 0.59 kg/BCM. Increased pattern size is used in the partially weathered material, while a decreased pattern size is utilised in harder material. The firing sequence blasts ore blocks separately from waste blocks. This style of blasting preserves the integrity of the ore/waste contacts to allow for visual identification of the zones by the pit geologists. The drill rigs are equipped with dip meters to measure accurately the drill-hole depth during drilling.

Ore is zoned in the geological modelling process to a minimum insitu width of 10 m by 5 m by 3.3 m, which reflects the minimum width that the current mining method and equipment can practically achieve.

Mine Infrastructure and Crusher

Ore material is crushed at Gounkoto with a primary Sandvik CJ815 jaw crusher and secondary cone Sandvik CS660 crusher. The final crushed product is minus 40 mm (80%) and minus 60 mm (20%) at a throughput rate depending on the final crushed size requirement. For the selected 35 mm product size the throughput rate is 360 tph.

The crushed material is stockpiled using an inclined slewing conveyor, allowing material classified on grade or oxidation to be separated and allowing loading and crushing to be occur simultaneously.

The ore is loaded from the crushed stockpile into haul trucks by the current haul contractor, SFTP. The haul fleet consists of 13 Volvo 50 t FMX tipper trucks.

The trucks transport ore from Gounkoto to Loulo, approximately 30 km away, using an average cycle time of 90 minutes. The trucks deliver crushed ore to the ROM pad at Loulo and feed the crushed ore directly into the hard rock circuit via the existing gyratory crusher.

Gounkoto has existing heavy mining equipment fleet maintenance and fuel storage facilities, as well as supporting offices and stores and a bulk fuel storage facility.

Equipment, Utilisation and Efficiencies

As of 31st December 2017, the Gounkoto contractor, DTP Terrassement, operates a fleet of thirty 90 t Caterpillar 777 trucks, and five hydraulic backhoe excavators. This fleet currently has a capacity to mine 35 Mtpa, as per the 2018 operational plan. The primary mining fleet is summarised in Table16-3.

18^th September 2018 Page 238

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-3 Gounkoto Mining Fleet as of 31^st December 2017

Equipment	Work Group	OEM	Unit	2018
Excavators	Production Unit	Liebherr	EX9350	5
		Liebherr	EX9250	1
		Liebherr	EX984	0
		Komatsu	PC1250	1
Support Excavators		Komatsu	PC800	1
		Komatsu	PC350	1
		Caterpillar	CAT390D	1
		Caterpillar	CAT336	1
Trucks	Production Unit	Komatsu	HD785	7
		Caterpillar	CAT 777D	14
		Caterpillar	CAT 777F	16
		Caterpillar	CAT 777G	1
	Production Unit	Atlas Copco	ROC-L8	2
		Atlas Copco	DM-30	9
Drill Rigs	RC	Atlas Copco	ROC-L8 RC	1
		Atlas Copco	SMARTROC60	3
		Atlas Copco	SMARTROC60	3
Water Cart	Support Unit	Caterpillar	CAT 773	3
Dozers	Support Unit	Caterpillar	CAT D9 - (Track)	9
		Caterpillar	CAT 834H - (Wheel)	2
Graders	Support Unit	Komatsu	GD850	1
		Caterpillar	CAT 16M	2
Wheel Loaders	Support Unit	Komatsu	WA470	1
		Caterpillar	CAT 988	1
		Caterpillar	CAT 988	1
		Caterpillar	CAT 966	2

The contractor is required to provide an availability of at least 80%, and the utilisation is 85%. Historically, the contractor has been able to provide over 80% availability. Both these values have been applied to productivity calculations for 2018 and beyond.

Dewatering

Gounkoto has two sumps, one in the southern section and one in the northern pit. Water is pumped to surface by a double lift arrangement. Pumps on the bank of the bottom sump stage pump to a higher sump (four pumps in the south, two in the north), where diesel pumps and electrical submersible pumps pump the water to surface via multiple high density polyethylene (HDPE) lines. During the wet season, pumping volumes average 5,000 m3/day from the southern sump and 4,000 m3/day from the northern sump. Gounkoto hasback-up diesel pumps available as spares.

Sixteende-watering bores have been drilled on the east side of the pit and five on the west side to actively lower the groundwater level. The vertical bores maintain groundwater level just below pit bottom; however, five horizontal drains have been drilled in the south wall to depth of 50 m to relieve pressure on this wall. Horizontal bores exhibit low flow rates and are monitored at least weekly; additional horizontal drains will be drilled if necessary. At this time the drains in place appear to be performing adequately. A total of 15 additional, deeper,de-watering holes have been planned with 13 already installed along with an additional 10 piezometers, to manage draw down rates.

18^th September 2018 Page 239

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Waste Dumps

The following design has been used where waste material is dumped on both the hanging wall and footwall side of the pit, at a maximum distance of 1 km from the ramp exit. Dumps are designed according to the following guidelines:

● Ramp widths of 30 m, except for drainage ramps which are 10 m

● Overall slope angle of 23º and batter angles of 37º

● Silt catchment provisions around the entire dump

● Maximum height of 50 m

● Small ramps and drainage gullies are provided where needed to providerun-off. All benches are gently sloping towards drainage areas

● Utilisation of topographical features to minimise the footprint, where possible

● Dumps drain away from the pit

The revised Super Pit design required some of the dumps in the west to be moved and the dumping method to be changed to accommodate an additional 230 Mt. Scheduling changes have allowed in pit backfilling to be planned, reducing the impact of the space constraint due to the Falémé river to the west of the pit. The following further guidelines were followed:

● A 30 m gap between toe of the dump and pit edge was maintained.

● A hardcore bund is required to allow water to drain away from the dump and stop waste slumping into the river.

● Increased dump capacity meant increased weight and pore water pressure that will require greater control as part of west wall slope stability management.

Life of Mine Plan

The Gounkoto mining plan has been sequenced in such a way that the south pit will be mined out in advance of the north pit. The schedule shows that the south pit will be mined out by April 2019. This will allow the waste mined from the north pit to be backfilled into the south pit allowing for a shorter haul than the conventional route of tramming out of the pit and onto a waste dump.

The additional ore mined in 2019, will be stockpiled to ensure the plant feed requirements are maintained during the waste pushback mining in years 2020 and 2022.

Table16-4 details the life of mine production plan for the Gounkoto pit.

18^th September 2018 Page 240

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-4 Life of Mine Production Plan for the Gounkoto Pit

Item	Units	LOM	2018	2019	2020	2021	2022	2023	2024
Open Pit Mining
Waste
Gounkoto	kt	200,921	33,098	32,099	37,837	36,344	28,250	26,692	6,600
Faraba	kt	8,954	-	-	-	-	2,985	2,985	2,985
Total Waste Tonnes	kt	209,875	33,098	32,099	37,837	36,344	31,235	29,677	9,585
Ore
Gounkoto	kt	13,660	3,337	2,749	1,041	1,544	919	1,632	2,438
Faraba	kt	2,480	-	-	-	-	827	827	827
Total Ore Tonnes	kt	16,140	3,337	2,749	1,041	1,544	1,746	2,458	3,265
Gold Grade
Gounkoto	g/t	5.03	3.61	5.88	4.84	3.97	4.82	3.88	7.62
Faraba	g/t	2.51	-	-	-	-	2.51	2.51	2.51
Total	g/t	4.64	3.61	5.88	4.84	3.97	3.73	3.42	6.32
Gold Ounces
Gounkoto	koz	2,209	387	520	162	197	142	204	597
Faraba	koz	200	-	-	-	-	67	67	67
Total	koz	2,409	387	520	162	197	209	270	664
Strip Ratio	w:o	13.0	9.9	11.7	36.4	23.5	17.9	12.1	2.9

18^th September 2018 Page 241

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

16.3   Loulo Underground Mining Operations

Introduction

The underground ore comes from Yalea underground mine, which was developed in 2008 as the Yalea pit came to the end of its life, and the Gara underground mine, which is a more recent extension (2009) of the former Gara open pit. During 2017, Yalea produced 1.5 Mt at 8.18g/t and Gara produced 1.2Mt at 3.75 g/t via an open stoping with backfill mining method.

Both mines have separate management teams and are run independently from each other, although equipment maintenance is undertaken in the Central Workshop which is shared. All mining is carried out as an owner operation.

The Yalea and Gara underground mines are accessed via portals located in open pits and a box cut. The 5.5 m wide by 6.0 m high declines were originally developed as twin declines with one decline carrying a conveyor belt system from the underground crusher to surface for both ore and waste handling purposes, and the other is used for vehicle access.

The lower parts of the mines have been developed as single declines with truck haulage up to the crushers which feed ore and waste onto the conveyors. Currently the crushers are located at 208 level in Yalea and 90 level in Gara. These are proposed to be the end points of the conveyors and all haulage will be diesel trucks to these crushers.

Most of the underground mining production equipment was replaced in 2016. The current fleet is detailed in Table16-5.

18^th September 2018 Page 242

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-5 Yalea and Gara Mining Fleet as of 31st December 2017

Equipment	OEM	Unit	2018
LHDs	Sandvik	LH621	11
Trucks	Sandvik	TH663	11
Development Drills	Sandvik	DD421	5
Production Drills	Sandvik	DL421	4

TheLow-Haul-Dump Loaders (LHDs) are equipped with remote control systems. Sandvik has a support team and warehouseon-site to support the operations. Some Caterpillar ancillary equipment is still in use.

Mine Development

In the lower parts of the both mines the declines have been split into multiple access ramps; below 230 level in Yalea and below 145 level in Gara. The ramps are spaced at approximately 500 m intervals along strike. This change has reduced the amount of waste development in the footwall, while limiting the loader tramming distances to mostly below 200 m. The development layout for Yalea and Gara is shown in Figure16-3 and Figure16-4 respectively.

Level accesses are at a vertical interval of 25 m in all new long hole stoping areas; with 20 m level intervals in stoping under rock fill (SURF) areas and some upper areas of the mines were developed with 20 m level interval and are still in production.

Primary development ends and cross cuts are 5.5 m wide by 6.0 m high, sized to fit the Sandvik 60 t trucks, while ore drives and supporting development are 5.0 m wide and 5.5 m high. Ground support is mainly split sets with mesh.

18^th September 2018 Page 243

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-3 Yalea Life of Mine Plan Development Long Section (Looking East)

Figure16-4 Gara Life of Mine Plan Development Long Section (Looking West)

Mining Methods

Long hole transverse open stoping and long hole longitudinal retreat open stoping are currently being used. Stoping under rock fill (SURF) mining is planned to be introduced in the weathered areas of Yalea south upper.

Long Hole Open Stoping

Current mining in the lower areas of the mine is an underhand (top down) long hole stoping method which retreats to central accesses in an echelon format. Panels are 50 m long, mined as single level stopes (25 m high) and filled with cemented paste fill. The paste fill is exposed by the mining of both the panel below and the next panel on the same level. At Yalea, 102 mm holes are drilled, while at Gara 89 mm holes are drilled in a fan arrangement and blasted as required to sustain production. Four to five panels are in production at any time. The ring design is modified using geological data to reduce dilution in each panel. Remote LHDs are used to load out blasted material.

Transverse long hole open stoping is used in the wider (over 15 m wide) parts of Yalea North. Transverse stopes are 15 m to 30 m along strike and mined from hanging wall to footwall. Figure16-5 and Figure16-6 illustrate the Yalea and Gara long hole stoping areas.

18^th September 2018 Page 244

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-5 Yalea Stoping Method Long Section

Figure16-6 Gara Stoping Method Long Section

Stoping Under Rock Fill (SURF)

The SURF method is proposed for the part of Yalea South Upper that has weathered transition or saprolite present in the mineralised zone, or the immediate hanging wall. The method will be used as it is expected to provide more consistent production in the poorer ground conditions observed in geotechnical drilling and geotechnical block models. Figure16-5 illustrates the Yalea SURF stoping areas.

The capital development design for SURF mining will support a change to long hole open stoping, if required, after mining of this area has commenced and the geotechnical performance of the ore zone and hanging wall is better understood.

SURF mining involves:

18^th September 2018 Page 245

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

● Filling the stopes with waste rock fill from tipping points at the top of the mining area to support the hanging wall.

● Tight blasting production drill rings (up holes) against the waste.

● Mucking ore until waste has rilled into the draw point.

● Tipping additional waste into the top of the mining area to provide support to the hanging wall.

From an operational perspective, SURF is very similar to longitudinal sub level caving (SLC). The exception is that in SURF, waste rock is tipped into the stope to fill mined voids and prevent caving of the hanging wall. In contrast, SLC works by the hanging wall caving to fill the mined void. In SLC, this caving of the hanging wall will often propagate through to the surface, causing subsidence. In SURF, the intention is to prevent this large-scale caving of the hanging wall.

The development design has waste tip points at the top of the SURF mining area at 20 m spacing to minimise the span of hanging wall exposed (Figure16-7). Production levels are developed with partial footwall drives to enable more competent long hole stoping areas to be mined first (Figure16-8).

Figure16-7 Stoping Under Rock Fill (SURF) Development Layout at Top of Mining Area (85 Level), Showing Waste Tipping Points

18^th September 2018 Page 246

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-8: Stoping Under Rock Fill (SURF) Development Layout of Production Level (185 Level)

Ventilation

At Yalea there are three declines that serve as intake airways: two in the Yalea and one atP-125, with a further intake in a chilled air raise at Yalea North. Two ventilation raises serve as return airways. Primary fans at Yalea are 630 kW units with two located at both the North raise borehole and the South raise borehole. Measured air flow at the North location is 277 m3/s and 273 m3/s at the South. Secondary ventilation to working areas is provided with 110 kW to 180 kW fans and 1,200 mm ventilation ducting.

Primary ventilation at Gara intakes through the twin portal and a chilled air raise bore; and exhausts through a pit vent access and return air raises to surface. Secondary ventilation to working areas is provided with 110 kW to 180 kW fans and 1,200 mm ventilation ducting. Backup power is available for the ventilation fans.

Large refrigeration plants have been constructed at both Yalea and Gara. These plants are identical and cool the intake air by 13°C with a throughput of 150 m3/s. This air enters the mine via the chilled air raise bores to 260 level and to 280 level from where it is mixed with the other fresh air. Both plants were commissioned in early 2017. Power usage of each plant is 3.2 MW.

Backfill

In 2015, paste fill plants were commissioned at both mines. The paste fill plants produce a mixed paste/aggregate fill. The Yalea paste plant produces 3,100 m3 per day and the Gara paste plant produces 2,750 m3 per day.

The ratio of slag/cement/tailings can be varied to change the strength of the backfill with the 90% tailings, 6% cement, and 4% slag giving the highest strength. The type of backfill required from either plant is based on geotechnical modelling of the opening to be filled.

The design strength of paste fill varies from 0.3 MPa to 0.5 MPa compressive strength for stopes that are not exposed below, that is an overhand (bottom up) stoping; to 0.9 MPa to 2.4 MPa

18^th September 2018 Page 247

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

compressive strength for stope that will be exposed below, that is an underhand (top down) stoping method. The plant is currently operating on 100% classified mill tailings, while previously, a 30% aggregate: 70% paste ratio, and a 50% aggregate: 50% paste ratio, have been used. The current configuration reduces pipe wear and the reduced viscosity enables paste reticulation to the extremities of the mine. Cement percentages vary from 2.2% to 5.7% for the above design strength range.

Initially, reticulation issues resulted from long horizontal lines and resulted in excessive pressure on the lines and limited the speed of pumping causing some paste line blockage. Raise boring has been used to drill additional lines; Gara now has two lines, and Yalea has three lines.

Dewatering

The current total pumping capacity at Gara is 120 l/s. The current reticulation uses a series of 20 l/s Mono pumps cascaded from one sump to another up to the main sumps at 135 level. From here it is either pumped to the 40 level XC8 pump station, where water is pumped direct to surface via three 110 mm lines or pumped via a backup system to surface using two 150 mm lines.

The current pumping rate at Yalea is 110 l/s from the underground and 40 l/s from the old open pit. The current reticulation uses the series of Mono pumps in the deeper areas of the North declines pumping at 20 l/s by daisy chain to main sumps at 35 level. From here it is pumped by larger pumps at 40 l/s to the main station on 258 level. From the 258 level sump it is pumped via a sump on 168 level to the main sump at 88 level. In the southern decline, water is pumped to a sump on 208 level from where it is pumped to the 88 level, the main pump chamber. From there water is pumped straight to surface via four 150 mm steel lines.

The electrical feed to the pump stations is supplied via a specific borehole carrying the cable or though the 11kV underground ring feed. There is an emergency power supply arrangement to these pumps.

Life of Mine Plan

The life of mine schedule for Yalea and Gara was created in Datamine 5DP / EPS software, a task-based dependency scheduler. The overall Loulo- Gounkoto complex life of mine was created in Excel by combining the output from the underground and open pit scheduling packages.

The Yalea life of mine extends to 2028 and Gara to 2032 (Table16-6 and Table16-7).

Total life of mine lateral development for Yalea is 39,000 m with capital development forming 44% of this at 17,200 m. The life of mine operating development, including ore drives, is 21,800 m.

Total life of mine lateral development for Gara is 45,000 m with capital development forming 47% of this at 21,300 m. The life of mine operating development, including ore drives, is 23,700 m.

18^th September 2018 Page 248

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The current fleet is expected to be sufficient and the total labour force will slowly reduce from the current total of 1,260 people.

18^th September 2018 Page 249

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-6 Yalea Life of Mine Production Schedule

	Units	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	LOM   Total
All Ore Tonnes	kt	1,429	1,439	1,450	1,446	1,455	1,453	1,453	1,372	963	524	301	13,285
All Ore Grade	g/t	6.25	6.22	5.92	6.94	6.08	5.31	5.42	5.07	4.31	4.80	3.94	5.70
All Ore Ounces	koz	287	288	276	322	285	248	253	224	133	81	38	2,435
Stope Diluted Tonnage	kt	1,252	1,262	1,273	1,329	1,316	1,365	1,453	1,372	963	524	301	12,411
Stope Diluted Grade	g/t	6.39	6.26	5.85	6.92	6.12	5.30	5.42	5.07	4.31	4.80	3.94	5.68
Ore Development Tonnage	kt	177	177	178	116	139	88	-	-	-	-	-	874
Ore Development Grade	g/t	5.29	5.92	6.44	7.15	5.73	5.46	-	-	-	-	-	5.99
North All Ore Tonnes	kt	581	408	265	349	240	245	159	172	229	78	0	2,727
North All Ore Grade	g/t	5.20	6.84	6.70	8.51	5.39	4.71	4.26	5.72	4.34	6.36		5.93
Central All Ore Tonnes	kt	611	441	249	25	-	-	-	-	-	18	0	1,346
Central All Ore Grade	g/t	7.51	6.05	4.47	3.26	-	-	-	-	-	2.93		6.32
South Upper All Ore Tonne	kt	19	33	129	237	131	471	669	798	447	271	301	3,506
South Upper All Ore Grade	g/t	3.43	4.47	5.18	5.17	4.86	4.26	4.61	4.61	3.75	4.04	3.94	4.41
South Lower All Ore Tonnes	kt	197	529	605	503	569	294	57	200	96	-	-	3,050
South Lower All Ore Grade	g/t	6.09	5.93	6.15	6.52	5.39	4.53	3.69	5.59	4.30	-	-	5.73
Far South All Ore Tonnes	kt	-	41	203	331	515	443	567	202	191	157	-	2,651
Far South All Ore Grade	g/t	-	5.84	6.47	7.46	7.49	7.28	6.86	5.81	5.58	5.57	-	6.83

18^th September 2018 Page 250

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-7 Gara Life of Mine Production Schedule

	Units	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032	LOM    Total
All Ore Tonnes	kt	1,181	1,251	1,250	1,250	1,253	1,252	1,255	1,254	1,249	1,252	1,157	917	396	223	200	15,338
All Ore Grade	g/t	3.90	3.72	4.16	4.18	3.89	3.97	4.86	4.39	4.24	4.22	4.45	4.53	4.4	4.53	3.36	4.20
All Ore Ounces	koz	148	150	167	168	156	160	196	177	170	170	166	133	56	33	22	2,071
Stope Diluted Tonnage	kt	970	1,057	1,046	1,008	1,012	1,001	998	1,254	1,249	1,252	1,157	917	396	223	200	13,741
Stope Diluted Grade	g/t	3.85	3.70	4.16	4.13	3.74	3.67	4.77	4.39	4.24	4.22	4.45	4.53	4.40	4.53	3.36	4.16
Ore Development Tonnage	kt	210	194	203	242	241	251	257	-	-	-	-	-	-	-	-	1,597
Ore Development Grade	g/t	4.12	3.84	4.15	4.38	4.50	5.16	5.19	-	-	-	-	-	-	-	-	4.52
North All Ore Tonnes	kt	292	277	238	299	419	559	371	187	105	-	-	-	-	-	41	2,788
North All Ore Grade	g/t	3.54	3.70	5.67	4.65	3.85	3.80	5.36	5.96	3.63	-	-	-	-	-	3.03	4.36
Central All Ore Tonnes	kt	444	356	349	298	311	79	-	13	17	-	-	-	-	-	-	1,867
Central All Ore Grade	g/t	4.15	3.82	3.90	3.27	3.11	2.57	-	4.41	3.71	-	-	-	-	-	-	3.66
South All Ore Tonne	kt	445	618	555	412	183	68	-	-	-	-	-	-	-	37	58	2,374
South All Ore Grade	g/t	3.88	3.67	3.61	4.42	3.98	3.84	-	-	-	-	-	-	-	4.68	3.03	3.85
South Upper All Ore Tonne	kt	-	-	2	84	110	208	189	181	92	50	-	-	-	-	-	915
South Upper All Ore Grade	g/t	-	-	2.36	4.62	4.67	3.65	4.48	3.64	3.47	3.02	-	-	-	-	-	3.98
Far South All Ore Tonnes	kt	-	-	107	157	228	148	411	478	688	805	530	147	44	-	76	3,821
Far South All Ore Grade	g/t	-	-	4.46	4.16	4.56	4.39	4.40	4.06	4.42	4.20	4.52	3.35	4.27	-	3.56	4.28
Far South Extended All Ore Tonnes	kt	-	-	-	-	-	190	284	394	348	397	626	769	352	187	26	3,573
Far South Extended All Ore Grade	g/t	-	-	-	-	-	5.10	5.13	4.38	4.30	4.42	4.39	4.75	4.41	4.50	4.01	4.56

18^th September 2018 Page 251

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

16.4 Gounkoto Underground Operations

Introduction

Underground mine operations for Gounkoto were determined in the December 2016 UndergroundPre-Feasibility. The mine life is six years. Ore production starts in 2025 and increases to a maximum of 525 kt in 2027, after which it steadily declines until end of mine life. The last year of production is in 2030 with 316 kt at 5.3 g/t Au.

The operation will include a combination of transverse and longitudinal open stopes and all material will be trucked to surface. There is only a single ramp. The exhaust system is a return air raise (RAR) will be developed simultaneously with the ramp.

Mine Development

The upper portions of the Gounkoto MZ1, MZ2, and MZ3 lodes are being mined in the Gounkoto open pit. The proposed decline will commence from within the pit to minimise the development distance to the orebody and eliminate the need to commence a decline in unconsolidated ground, which is typically 25 m below the surface and can occur up to 50 m below the surface.

The decline will start in 2025 while open pit mining is still underway. There will be an opportunity to stabilise the pit wall above the portal and return air adit positions before they are reached and required for mining. Figure16-9 is an isometric view of the underground mine below the pit and shows the location of the portal access and return air adit.

Figure16-9 Isometric View of Underground Design in Relation to Super Pit Design

18^th September 2018 Page 252

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The decline has been designed to access the orebody from the western side. This avoids accesses going through the Fault Gouge which is adjacent to the eastern side of the orebody and is predominantly classed as ‘Poor’ ground. The West Dipping Fault approaches the western side of the orebody in the upper part of the underground mining area and dips away from the orebody with depth.

The decline maintains a minimum 50 m horizontal clearance to the mining footwall contact of the orebody. The decline has been designed in an oval configuration with straights between curves to better suit truck haulage. Each loop in the decline has been designed to drop down 40 m vertically.

Figure16-10 is a longitudinal projection looking east showing the decline and level development.

Figure16-10: Longitudinal Projection Looking East Showing Decline and Level Development

Each level has a single access from the decline to the footwall drive. The accesses are designed to be central to the north-south extents of the orebody, but this is limited by the design of the decline which has a loop every second level.

18^th September 2018 Page 253

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-11 is a view looking east of a typical level design showing the variation in the position of the level.

Figure16-11 View Looking East of Typical Level Designs Showing Variation in Position of Level

A plan view of a typical level layout is shown in Figure16-12.

Figure 16-12 Plan View of Typical Level Layout

The levels are designed to have a footwall drive with crosscuts to the orebody for transverse stopes and access to ore drives for longitudinal stopes. The footwall drives on each level extend

18^th September 2018 Page 254

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

from the northern extents to the southern extents of the orebody and provide access to stope crosscuts and return airway systems.

The primary access and cross cuts are 6 m high by 5.5 m wide while ore drives are 5 m high by 5 m wide.

Mining Methods

Long hole transverse stoping and long hole longitudinal stoping will be used at the Gounkoto underground mine.

The extraction sequence for the wider parts of the orebody is a primary/secondary stoping sequence and requires a footwall drive to provide access to the secondary stope after a primary stope is extracted. The primary/secondary stoping sequence enables the filling secondary stopes withun-cemented fill which is a substantial cost saving.

In the transverse stoping area, ore crosscuts are driven from the footwall drive east towards and through the orebody to meet a hanging wall ore drive (slot drive) on the eastern side of the orebody.

Longitudinal stoping areas have a single ore drive driven along the centre or near to the hanging wall of the orebody. The drive is developed to the extents of the stoping block and stopes are taken, retreating to the access crosscut.

The long hole stoping method is a‘bottom-up’ mining method. The number of levels in each stoping block varies depending on the geometry of the orebody. Areas that have less ore or lower grade ore present opportunities for a crown pillar, where ore recoveries are lower. Figure16-13 shows a longitudinal projection of the grades of stopes. The-440 Level has been selected as the base of a block so that higher grade ore can be mined earlier. That block then goes up three levels to the base of the next block at-360 Level. The stoping blocks and stope types are shown in Figure16-14 and Figure16-15.

For longitudinal and transverse stopes, a burden of 2.3 m and a spacing of 3.3 m has been assumed using 89 mm diameter holes.

18^th September 2018 Page 255

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-13 Longitudinal Projection Looking East Showing Stopes at an Average Au Grade

Figure16-14 Longitudinal Projection Looking East Showing Stope Blocks by Mining Type

18^th September 2018 Page 256

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-15 Longitudinal Projection Looking West Showing Stope Blocks and Stope Types

Transverse Stopes

The transverse stopes are mined using a ‘primary/secondary’ sequence. This is a sequence where every second stope on a level is mined and filled with cement aggregate fill (CAF) or cement rock fill (CRF) before the remaining stopes in between are extracted and can be filled with loose rock fill.

All stopes on the sill level are to be filled with CAF to provide a stable crown for improved recovery when mining the blind uphole stopes underneath. The stopes that are above the sill levels are to be filled using the conventional primary/secondary method.

Where horizontal stope widths, across strike, are greater than approximately 15 m, the stopes have been designed as transverse long hole stopes. This is generally in the MZ3 zone which is in the northern part of the mine. Stope widths less than 15 m have been designed as longitudinal long hole stopes. Figure16-16 is a longitudinal projection of a transverse stope layout.

18^th September 2018 Page 257

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-16 Longitudinal Projection of Transverse Stope Layout

Longitudinal Stopes

Longitudinal long hole stopes are designed for mining the narrower parts of the orebody. The stopes are mined from the furthest extent of the orebody and retreated back to the access. A single ore drive is used for access for drilling and mucking. Stoping is commenced with a slot at the extreme end of the stope and the stope can be retreated back to a maximum stable span which, in this case, is designed to be 20 m. The stope is then initially filled with CRF to form a solid pillar for the next stope and then finished by filling with loose rock. The next stope is then commenced by slotting against the CRF. A longitudinal projection of the longitudinal stoping method is shown in Figure16-17.

18^th September 2018 Page 258

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-17 Longitudinal Projection of Longitudinal Longhole Stoping

Ventilation

The return airway raise will be accessed from an adit within the final pit to minimise raising metres, eliminate the need for apre-sink, and to minimise underground development to the bottom of the raise.

The surface primary exhaust fans will be located on the exhaust portal bench. The primary fan duty is 180 m³/s at 1.15 kg/m³ at 1,625 Pa collar total pressure. The fan will be approximately 400 kW and will consist of two fans in parallel.

Based on experience at Gara and Yalea, Gounkoto underground will require some refrigeration to achieve the Randgold standard of not having persons working outside ofair-conditioned cabins at temperatures above 30° wet bulb. Air will be chilled by a Bulk Air Cooler connected to a fridge plant. The fridge plant is currently assumed to be similar to those used at Yalea and Gara, with further detailing to be completed in the future. Chilled air will be introduced into the main portal via a nearby second portal off the same bench in the open pit, then will travel down the ramp system. Each level will be ventilated in series off the ramp and the air then enters the bottommost RAR connection.

18^th September 2018 Page 259

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Surface and Underground Infrastructure

There are four underground electrical substations designed in the mine. The electrical substations have been designed to each service three levels. The lower most substation is located in the decline above the lower pump station.

A permanent pump station has been designed at the bottom of the underground workings. It is located near the end of the decline and will pump to the elevation of the portal and discharge into the pit workings. Pumped water will be routed via pipes situated in service drives located off the decline.

CAF is to be used in all sill stopes. The Euromix Continuous Mobile CAF plant, designed by Nisbau GMBH, is currently located at the Loulo mine and is to be relocated to a location on the western side of the Gounkoto open pit and will prepare CAF for delivery underground via service hole at this location.

The slurry plant is to be located on the western side of the pit alongside the CAF plant and will prepare slurry for delivery underground via a service hole at this location.

This approach is similar to what is used at Gara and Yalea, with adjustments expected as production begins.

Life of Mine Plan

The Gounkoto underground life of mine plan is based the productivity rates and time rates achieved at Gara and Yalea, as shown in Table16-8 and Table16-9.

Table16-8 Productivity Rate

Resource	Rate	Units
Dev Jumbo	250	metres/month
ITH Rig	20	metres/day
Cable Bolter	190	Cable bolt metres/day
Long Hole Rig	5,700	Production drill metres/month
Production Bogger	1,000	insitu tonnes/day
Development Bogger	12	metres/day
Backfill Bogger Rockfill	1,000	rockfill tonnes/day
Backfill Bogger Cemented Fill	1,000	cemented fill tonnes/day
Cubex V30	2	metres/day

18^th September 2018 Page 260

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-9 Time Rates for Activities

Activity	Rate	Units
Development Default	4	metres/day
Decline Development	4	metres/day
Production Development	3	metres/day
Production drilling	5,700	prod drill metres/month
Cable Drilling	190	cable bolt metres/day
Service Hole Drilling	50	metres/day
Raise Bore Hole	3.5	metres/day
Return Air Raise	2	metres/day
Slot Rise	2	metres/day
Long Hole Stope Production	1,000	insitu tonnes/day
Sill Prep	2	days
Stope Prep	7	days
Fill Preparation	3	days
Fill	800	rock fill tonnes/day

The production schedule and key mine physicals are summarised in Table16-10.

Table16-10 Gounkoto Production Schedule and Key Mine Physicals

Physical	Unit	2025	2026	2027	2028	2029	2030	LOM
Lateral Development	m	4,278	5,900	4,801	422	-	-	15,401
Vertical Development	m	204	375	665	508	460	271	2,483
Ore Tonnes	kt	42	348	525	468	418	316	2,117
Contained Au	k	9.04	83.45	97.98	92.47	80.07	58.33	421.33
Grade	g/t	6.6	7.5	5.8	6.1	6.0	5.7	6.19
Waste Tonnes	kt	310	325	247	9	-	-	891
Backfill	kt	-	103	131	201	184	233	853
Cemented Rock Fill	kt	-	59	87	79	47	99	372
Cemented Aggregate Fill	kt	-	26	18	94	93	36	268
Rockfill	kt	-	18	26	28	44	97	213
Cable-bolt Metres	km	5	19	31	22	19	10	106
Production Drill Metres	km	-	22	44	49	46	35	197

The mine will be fully developed by the end of 2028. Schedule optimisation software was used, which has led to planning a high lateral capital development rate early in the mine life, which will permit the establishment of higher-grade stopes earlier than usual. Despite the extra costs incurred, a higher NPV will be achieved.

Total life of mine lateral development is 15,401 m with capital development forming 45% of this at 6,987 m. The life of mine operating development, including ore drives, is 8,414 m. On average the operating waste component constitutes approximately 42% of the total operating development metres. Total life of mine vertical development is 2,484 m with capital development forming approximately 16% of this at 388 m. Production slot raise development makes up the remaining 84% or 2,096 m. The annual development is summarised in Figure16-18.

18^th September 2018 Page 261

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Figure16-18 Development Coloured by Year

The location of the annual ore production from the underground mine is shown in Figure16-19. All modifying factors such as stope dilution and mining recoveries have been applied.

Figure16-19 Stoping Coloured by Year

The total ore production for thelife-of-mine is 2.1 Mt.

The annual production reaches a peak in 2027 when the decline has reached its maximum depth and the greatest number of transverse stopes are online. Post 2027 ore development is virtually nil, and the mine becomes a pure stoping operation. Production then steadily declines until end of mine life. The drop off in the production tonnage in these latter years is considered unavoidable due to the geometry and tonnages left in the remaining stopes.

The life of mine gold production is 421 koz with an average life of mine grade of 6.2 g/t Au.

18^th September 2018 Page 262

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The head grade for the initial years of the life of the mine is higher due to targeted scheduling identified during schedule optimisation. Higher grade development and stopes were scheduled earlier on, while at the same time, keeping the mine life limited so as to maximise NPV. For the years from 2028 to 2030 the grade remains fairly constant, however, a consistent drop in total ore tonnage leads to an equally consistent fall in ounces for the last three years.

Overall this mine schedule is considered the best approach for economic efficiency and maximising pay back opportunity whilst still maintaining a technically sound mining sequence.

Manpower

Total contractor and owner manning will reach a peak of 233 inmid-2027. There is a steady build up in manning from 81 at projectstart-up in 2025 until full production in 2027. The manning remains around 200 over the peak production years 2026 and 2027 before steadily dropping in conjunction with production levels. The contractor manning is approximately 195 and the owner manning ranges between 16 and 21. Labour will be moved across from the current Gara and Yalea operations.

16.5 Loulo-Gounkoto Life of Mine Plan

The Loulo Life of Mine has the operation mining a total of 52 Mt of ore at 4.73 g/t until the year 2032 (Table16-11). This gives Loulo a remaining life of 15 years. During this15-year period the total plant ore feed tonnage will amount to 56 Mt, including current stockpiles, at an average feed grade of 4.55 g/t resulting in a recovered 7.5 Moz at an average processing recovery of 92%. The Yalea operation will continue until 2030 and Gara until 2032, with the Loulo Open Pits mined from 2024 through to 2027. The Gounkoto Open Pit will be mined out in 2024 with the Faraba Pit starting in 2022 and ending in 2024. The Gounkoto Underground Operation will commence in 2025 and will continue until 2030.

A total of 28.6 Mt of ore will be mined from the Yalea and Gara underground operations with a further 5.4 Mt mined from the Loulo 3, Baboto, and Gara West open pits, at the Loulo Operations. The Gounkoto Operations will contribute 16.1 Mt mined ore tonnes from the Gounkoto and Faraba open pits with the Gounkoto underground operation adding 2.1 Mt.

The timing of the Gounkoto underground operations aims to sustain total underground production as the Yalea operation tails off. This keeps the fleet and labour demand level over this period. No extra labour will be required, and the main fleet will peak at 13 trucks, 10 LHDs, four development rigs and four production rigs.

Similarly, the timing of the various open pits aims to sustain open pit production after the Gounkoto Super Pit tails off.

18^th September 2018 Page 263

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table16-11 Loulo-Gounkoto Life of Mine Plan

	Units	LOM	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032
Waste (Mt)
Loulo 3	Mt	45.9	-	-	-	-	-	-	19.2	26.7	-	-	-	-	-	-	-
Baboto	Mt	9.5	0.6	-	-	-	-	-	-	-	-	8.9	-	-	-	-	-
Gara West	Mt	7.4	-	-	-	-	-	-	-	-	7.4	-	-	-	-	-	-
Gounkoto	Mt	200.9	33.1	32.1	37.8	36.3	28.3	26.7	6.6	-	-	-	-	-	-	-	-
Faraba	Mt	9.0	-	-	-	-	3.0	3.0	3.0	-	-	-	-	-	-	-	-
Total Open Pit Waste	Mt	272.7	33.7	32.1	37.8	36.3	31.2	29.7	28.7	26.7	7.4	8.9	-	-	-	-	-
Gara U/G	Mt	2.1	0.3	0.4	0.3	0.3	0.3	0.3	0.2	-	-	-	-	-	-	-	-
Yalea U/G	Mt	1.4	0.3	0.3	0.2	0.3	0.2	0.1	-	-	-	-	-	-	-	-	-
Gounkoto U/G	Mt	0.9	-	-	-	-	-	-	-	0.3	0.3	0.2	0.0
Total U/G Waste	Mt	4.4	0.7	0.6	0.6	0.6	0.5	0.4	0.2	0.3	0.3	0.2	0.0
Total Waste Mined	Mt	277.0	34.3	32.7	38.4	36.9	31.7	30.1	28.9	27.0	7.7	9.2	0.0
Ore (Mt)
Loulo 3	Mt	2.2	-	-	-	-	-	-	1.0	1.2	-	-	-	-	-	-	-
Baboto	Mt	2.3	0.3	-	-	-	-	-	-	-		2.0	-	-	-	-	-
Gara West	Mt	0.9	-	-	-	-	-	-	-	-	0.9	-	-	-	-	-	-
Gounkoto	Mt	13.7	3.3	2.7	1.0	1.5	0.9	1.6	2.4	-	-	-	-	-	-	-	-
Faraba	Mt	2.5	0.0	0.0	0.0	0.0	0.8	0.8	0.8	-	-	-	-	-	-	-	-
Total Open Pit Ore	Mt	21.5	3.7	2.7	1.0	1.5	1.7	2.5	4.2	1.2	0.9	2.0	-	-	-	-	-
Gara U/G	Mt	15.3	1.2	1.3	1.2	1.2	1.3	1.3	1.3	1.3	1.2	1.3	1.2	0.9	0.4	0.2	0.2
Yalea U/G	Mt	13.3	1.4	1.4	1.5	1.4	1.5	1.5	1.5	1.4	1.0	0.5	0.3	-	-	-	-
Gounkoto U/G	Mt	2.1	-	-	-	-	-	-	-	-	0.3	0.5	0.5	0.4	0.3	-	-
Total U/G Ore	Mt	30.7	2.6	2.7	2.7	2.7	2.7	2.7	2.7	2.7	2.6	2.3	1.9	1.3	0.7	0.2	0.2
Total Ore Mined	Mt	52.3	6.3	5.4	3.7	4.2	4.5	5.2	6.9	3.9	3.4	4.3	1.9	1.3	0.7	0.2	0.2
Total Ore to Plant	Mt	55.7	5.1	4.7	4.7	4.7	4.7	4.7	4.7	4.7	4.5	4.7	4.7	1.9	1.3	0.2	0.2
Grade (g/t)
Loulo 3	g/t Au	4.99	-	-	-	-	-	-	3.75	5.97	-	-	-	-	-	-	-
Baboto	g/t Au	2.79	2.64	-	-	-	-	-	-	-		2.82	-	-	-	-	-
Gara West	g/t Au	2.62	-	-	-	-	-	-	-	-	2.62	-	-	-	-	-	-
Gounkoto	g/t Au	5.03	3.61	5.88	4.84	3.97	4.82	3.88	7.62	-	-	-	-	-	-	-	-
Faraba	g/t Au	2.51	-	-	-	-	2.51	2.51	2.51	-	-	-	-	-	-	-	-
Total Open Pit Ore Mined	g/t Au	4.40	3.52	5.88	4.84	3.97	3.73	3.42	5.73	5.97	2.62	2.82	-	-	-	-	-
Gara U/G	g/t Au	4.16	3.90	3.72	4.16	4.09	3.89	3.94	4.86	4.36	4.13	4.22	4.22	4.50	4.40	4.53	3.36
Yalea U/G	g/t Au	5.70	6.25	6.22	5.92	6.94	6.08	5.31	5.42	5.07	4.31	4.80	3.94	-	-	-	-
Gounkoto U/G	g/t Au	6.19	-	-	-	-	-	-	-	6.60	7.50	5.80	6.10	6.00	5.70	-	-
Total U/G Ore Tonnes Mined	g/t Au	4.97	5.19	5.06	5.10	5.61	5.07	4.68	5.16	4.76	4.66	4.72	4.63	4.97	4.98	4.53	3.36

18^th September 2018 Page 264

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

	Units	LOM	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032
Total Tonnes to Plant	Mt	4.73	4.21	5.47	5.03	5.01	4.54	4.08	5.51	5.14	4.14	3.84	4.63	4.97	4.98	4.53	3.36
Total Grade to Plant	g/t Au	4.55	4.74	5.07	5.05	5.08	4.32	4.30	4.37	5.50	4.92	4.03	3.05	4.14	4.05	4.53	3.36
Ounces (koz)
Loulo 3	koz	353	-	-	-	-	-	-	118	235	-	-	-	-	-	-	-
Baboto	koz	208	29	-	-	-	-	-	-	-		179	-	-	-	-	-
Gara West	koz	73	-	-	-	-	-	-	-	-	73	-	-	-	-	-	-
Gounkoto	koz	2,209	387	520	162	197	142	204	597	-	-	-	-	-	-	-	-
Faraba	koz	200	-	-	-	-	67	67	67	-	-	-	-	-	-	-	-
Open Pit Ounces Mined	koz	3,042	416	520	162	197	209	270	782	235	73	179	-	-	-	-	-
Gara U/G	koz	2,051	148	150	167	164	157	159	196	176	166	170	157	133	56	33	22
Yalea U/G	koz	2,435	287	288	276	322	285	248	253	224	133	81	38	-	-	-	-
Gounkoto U/G	koz	421	-	-	-	-	-	-	-	9	84	98	92	81	58	-	-
U/G Ounces Mined	koz	4,907	435	437	443	487	441	407	449	408	383	349	287	213	114	33	22
Total Ounces Mined	koz	7,950	851	957	605	683	650	677	1,231	643	456	528	287	213	114	33	22
Total Ounces to Plant	koz	8,146	775	768	767	769	654	651	663	836	716	611	463	259	163	33	22
Recovered Ounces	koz	7,522	715	711	711	711	601	601	613	768	660	564	428	241	148	30	20

18^th September 2018 Page 265

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

17 Recovery Methods

17.1 Processing Plant

The Loulo processing plant uses a conventional carbon in leach (CIL) gold extraction process with a budget throughput in 2017 of 4.8 Mt, which is in line with the plant design capacity. A simplified metallurgical flowsheet of the process can be found in Figure17-1.

Figure17-1 Simplified Metallurgical Flowsheet

Since 2014 multiple optimisation projects have been undertaken, resulting in increased throughput and improved recoveries. It is expected that the improvement initiatives will continue but are considered part of operations and not necessarily confined to capital projects. The expected throughputs and recoveries are based on historical metallurgical testwork and the actual operational performance.

The upgraded process plant remains a conventional crushing, milling, gravity, CIL, and tailings disposal circuit. It will process an average of 580 tph using the following circuits:

● Crushing – three stage crushing for the hard rock sulphide ores and a single stage roll toothed crusher for the soft weathered oxide ores.

● Milling – one primary mill (7 MW); two identical single stage ball mills (4.5 MW), one scat conveyor system is inserted to return all mills scats back to primary mill via a cone crusher.

18^th September 2018 Page 266

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

● Gravity - four centrifugal primary concentrators followed by two intensive leach reactors to treat primary concentrates.

● CIL recovery process.

● Zadra elution process.

● Electrowinning.

● Tailings pumping/deposition split between slime dam and paste plant.

Processing plant production statistics are in Table17-1.

Table17-1 Processing Plant Production Statistics

	2015 Actual	2016 Actual	2017 Actual	2018 Planned
Total Ore to Plant (kt)	4,543	4,875	4,918	5,081
Recovery (%)	90.32	91.00	92.70	92.25
Recovered ounces	630,167	707,116	177,565	715,042

Crushing – Hard Ore

Hard rock is delivered to the tipping bin. Tipping can be completed from two directions.

The hard rock crusher plant consists of a primary gyrator crusher (FFE Minerals 1,300 mm by 1,750 mm) driven by a 450 kW electric motor. A tramp iron magnet is installed over the product conveyor to remove steel. A metal detector trips the conveyor if metal is detected.

The primary crusher operated at a rate of 700 tph, feeds the secondary crushers which operate in open circuit and then the tertiary crushers which operate in closed circuit. An additional metal detector is installed on the feed to the tertiary crushers.

The secondary circuit consists of two Sandvik CS6800 Hydrocone crushers and the tertiary circuit consists of four H6800 Hydrocone crushing units, all are powered by a 315 kW electric motor. The secondary and tertiary crushers are housed in separate buildings connected via transfer conveyors.

The final product from the tertiary crushers discharges onto a reclaim ore stockpile, via a conveyor system, and has a live capacity of about 40,000 t and a total of approximately 115,000 t. This is the feed to the SAG mill.

Crushing - Soft Ore

A separate facility exists to crush the soft ore when it is available. The soft ore crushing consists of a tooth roll crusher (MMD 625 - 3 tooth) driven by a 250 kW electric motor. This feeds a separate stockpile area and the roll crusher product is fed directly onto the ball mill conveyor feed.

18^th September 2018 Page 267

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Grinding and Classification

The milling section produces a leach feed with approximately 75% passing 75 microns. The milling circuit comprises of a SAG mill operating as the primary mill (6.1 m Ǿ and 9.5 m EGL and installed motor power of 7,000 kW to be upgraded to 8,000 kW) operating in open circuit. The oversize or scats arere-handled to the stockpile andre-crushed. The primary mill discharge feeds two parallel single stage overflow type ball mills (5.5 m Ǿ by 8.0 m EGL) with a power rating of 4,500 kW each, operating in closed circuit with a dedicated cluster of twelve 250 mm hydro-cyclones of which eight are typically in use.

An additional High Pressure Grinding Roll (HPGR) has been installed to treat scats at a rate of 100 t/h.

Gravity Concentration

A bleed stream from each cyclone cluster underflow is diverted to its respective gravity concentration circuit.

In each primary concentration circuit, the feed slurry from the cyclone underflow is fed on a horizontal vibrating screen to remove material greater than 2 mm from the concentrator feed that would otherwise plug the concentrator bowl riffles.

Screen undersize from each cluster feeds KnelsonXD-48 andQS-40 gravity gold concentrators after which the concentrates pass to a common feed hopper and Gemini dressing table located in the gold room. There is also the presence of an ILR1000 and ACACIA CS2000.

A recent initiative has been to test the effectiveness of the Knelson concentrators by removing them from the circuit. The water balance necessitated their feed to be sourced from recycled process water rather than fresh or raw water which resulted in significant corrosion of both the units but also their supporting structure located within the mill section. Indications are that due to there being a very low gravity recovery portion, that no deleterious effect has been witnessed.

Leaching and Adsorption

The CIL circuit consists of fourteen tanks that operate in series, each having a nominal capacity of 2,500 m3 giving a retention time of approximately 40 hours. Cyclone overflow, at a density of 35% solids, is introduced onto a linear trash screen (0.7 mm by 0.7 mm apertures) and the undersize goes to the leach feed thickener, the thickener underflow at a density of 50% presented as leach feed. Twopre-oxidation tanks receive the CIL circuit, each with six high-shear reactors (AachenREA-400). Cyanide is added into Tank 2 and doses automatically to control concentration around the desired value with ± 2%. An oxygen dispersion system is installed to maintain optimal dissolved oxygen across CIL to assist leaching consisting of power mixers (EKATO in front tanks #3 to 7). The oxygen plant, operated by Air Liquide, supplies 44 tpd. Hydrogen peroxide is injected at the thickener underflow, as required, to maintain the required dissolved oxygen levels necessary for the process. Each CIL tank is fitted with a mechanically swept cylindrical inter-tank screen (0.8 mm) complete with pumping mechanism.

18^th September 2018 Page 268

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Fresh carbon is introduced to tank fourteen and is advanced to tank three in counter current flow to the pulp, using recessed impeller vertical pumps.

Elution and Gold Recovery

Loaded carbon is recovered from CIL tank 3 into the acid wash cone. After elutriation, it is acid-washed using a 3% hydrochloric acid solution, followed by a caustic neutralisation/water flushing step. The rinsed carbon is transferred to one of two elution columns where the caustic/cyanide solution is circulated at elevated temperature (135°C) and pressures using the Zadra process. Loulo carbon stripping consists of two parallel circuits from harvest - elution column – heater and exchanger to electro-winning cells at the gold room.

The pressure Zadra method utilises a pressure strip vessel that reverses the chemical equilibrium of the adsorbed gold-cyanide complex on the activated carbon resulting in desorption of the gold-cyanide complex from the activated carbon into the strip solution.

The pressure Zadra process is conducted in abatch-by-batch process and requires approximately eight to 16 hours to complete.

The gold is then recovered downstream from the strip column by electro-winning. Six electro-winning cells are provided in the gold house for the electro-winning circuit where the pregnant electrolyte is introduced for gold deposition.

After each elution, the gold loaded stainless steel mesh cathodes are removed from the electro-winning cells and hosed down with high pressure water stream. This removes the plated gold onto a hopper where it is collected, settled, decanted and the sludge smelted after it has been dried in one of two electrically heated calcine ovens to produce doré bullion.

The barren carbon is now transferred to the adsorption circuit or to the carbon regeneration kiln where it is regenerated at C 700°C in a horizontal gas fired kiln.

The regenerated carbon is charged back into the CIL circuit.

Tailings Thickening and Paste Preparation

Tailings, discharging from No.14 CIL tank, gravitate through the tails linear screen (0.8 mm by 0.8 mm), then feed into the tails tank from where it is pumped to the Intermediate Plant. This is a technical requirement for the underground paste plants, which must use a detoxified coarse tailings as backfill material to fill the stopes. The Intermediate Plant has cyanide destruction together with arsenic fixing andtwo-stage cycloning to remove clay and fines (if present) and discharges the coarse fraction into a tank. The coarse fines are used as tailings and the fine slimes are pumped to the TSF. The tails pump station is equipped with two streams of four stage pumps and a flow diversion valve, where the delivery line of the duty (stainless steel pipe) is dedicated to high throughput deposition. At the valve station, there is a possibility to use a standby line (HDPE pipe) for low throughput operation. Stainless steel pipe rotation is planned to extend the life of main delivery line.

18^th September 2018 Page 269

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The tailings pipeline is a 450 mm diameter steel pipe, lying above ground and inside a trench, along a vehicle access road between the TSF and the processing plant. The return water line is in the same trench and consists of two 450 mm diameter HDPE pipe.

Recycled water is received from underground paste plants, and tailings storage facilities. Total recycled water is targeted at 80. Provision is made to allow intermittent discharge via the TSF detox pond, after chemical treatment.

Reagents

A dedicated reagent section has been provided for the warehousing of raw materials,make-up, and storage of reagents. These are:

● Sodium cyanide - delivered as solid briquettes in bulk bags packed within wooden crates. Make up involves hoisting the bags and lowering them onto a bag splitter above a makeup tank. After agitation for a short period, this solution is transferred to a dosing tank from where it is pumped to the addition points in the plant.

● Lime is delivered in powdered form also in bulk bags. It is made into slurry through a similar make up system to the cyanide and pumped to the plant addition points.

● Flocculants are delivered in powder form and made up in a traditional eductor hydration system.

● Caustic soda is dissolved in an agitated tank and pumped to the elution section.

● Hydrochloric acid is delivered in concentrated form which is stored in a tank and pumped to the elution section.

● Oxygen is produced by Air Liquide under contract via six PSA units outside the plant producing 44 tpd which is piped to the leach section.

● Hydrogen peroxide is stored in 1 m³ plastic isotainers.

These are mixed in a dedicated area outside the security fence and then pumped in as required. This minimises the number of vehicles that require access to the plant. Apart from the acid, the reagents are all in solid form when delivered and stored onsite.

Maintenance

The mineral processing operations have a formal scheduled maintenance system. Calendar based maintenance is conducted daily, weekly, two weekly, quarterly, six monthly, and yearly. The plant shuts for 14 hours every month when all mills stop and then twice a month one ball mill is shut down for eight hours. Statutory inspections and Original Equipment Manufacturer (OEM) checks are conducted at OEM recommended intervals based on the vendor and equipment. Mills undergo inspections yearly, Putzmeister pumps are inspected by OEM every six months, crushers are inspected yearly, by Sandvik, etc.

A comprehensive Supervisory Control And Data Acquisition (SCADA) on the equipment. For example, there is online, continuous monitoring of the mills for temperature (bearings and lubrication system), and vibration.

18^th September 2018 Page 270

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Oil testing is undertaken by Shell. Other condition monitoring includes thermography, relay testing, and vibration analysis. While contractors have been and continue to perform many of the Regulatory Compliance Mark testing, an group has been established and technicians received training in South Africa for thermography and are proficient in vibration and acceleration testing.

A total of 112 employees provide staffing of the maintenance department for processing on all shifts and rotations. Contractors and OEM representatives are utilised, as well, for specific tasks. Pragma is utilised as the maintenance software programme for planning, creation of job cards (work orders), and creating/storing historical maintenance and performance data. In addition, the team is starting to do failure analysis to better understand breakdown root causes and to see if a change in maintenance tactics will reduce downtime.

Corrosion

Although Loulo is located in a humid area with high rainfall, the fact that this is a relatively new plant (c. ten years) means that corrosion is still negligible, although it was necessary a few years ago to replace the acid tank with a rubber lined alternative. Following thickness testing of tanks in 2015, Loulo installed fibre reinforcements at the bottom of the CIL tanks.

Most areas in the Plant have been concreted, and spillage is reasonably well managed.

Plant MCCs

The Motor Control Centres (MCCs) are converted shipping containers fitted with passive point fire detection which is monitored locally and remotely at the control room, They are raised andair-conditioned to ensure optimal operating temperatures.

Cables are all bottom entry from the void below. All high voltage (HV) cabling and connections are thermo-graphically checked by a third-party ahead of the rainy season.

Mill motors are fed directly at 11 kV from the power station, via an underground pipe.

17.2 Production History

The Loulo processing plant has been operating since 2005. The production history is summarised in Table17-2 and Figure17-2.

18^th September 2018 Page 271

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table17-2 Loulo Processing Plant Production History

Year	Total		2005		2006		2007		2008		2009		2010
Loulo
Milled Tonnes kt		31,900		527		2,595		2,654		2,721		2,947		3,158
Gold Produced, koz		3,819		63		242		265		258		352		317
Gounkoto
Milled Tonnes kt		13,769
Gold Produced, koz		1,814
Loulo-Gounkoto
Milled Tonnes kt		45,668		527		2,595		2,654		2,721		2,947		3,158
Gold Produced, koz		5,633		63		242		265		258		352		317
Overall Recovery %		90.7		95.9		93.9		93.1		91.5		87.7		92.5
Year	2011		2012		2013		2014		2015		2016		2017
Loulo
Milled Tonnes kt		2,670		1,837		2,422		2,711		2,570		2,587		2,500
Gold Produced, koz		208		220		307		380		351		420		437
Gounkoto
Milled Tonnes kt		949		2,518		2,041		1,699		1,973		2,288		2,300
Gold Produced, koz		138		283		273		259		280		287		293
Loulo-Gounkoto
Milled Tonnes kt		3,619		4,354		4,463		4,410		4,543		4,875		4,800
Gold Produced, koz		346		503		580		639		630		707		730
Overall Recovery %		88.1		89.3		89.2		90.2		90.0		91.0		92.7

Figure17-2 Loulo-Gounkoto Processing Milled Tonnes by Ore Source

18^th September 2018 Page 272

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo-Gounkoto has demonstrated successful operation both in terms of processing throughput and in particular with gold recovery, demonstrating operational evidence for the ability to deliver an average of 4.8 Mtpa over the LOM.

17.3 Capital Projects and Plant Upgrades

Due to the competency of the Gounkoto ore, the Loulo plant circuit was upgraded with the addition of a primary ball mill in an open circuit configuration with a dedicated mill feed stockpile with 10 to 12 hours of live capacity which enables the consistent operation of the ball mill. The primary mill feed stockpile is sized to maintain steady state feed conditions to the primary ball mill. The incorporation of belt feeders underneath the stockpile alleviates material handling problems associated with sticky and fine material on the mill feed stockpile. The maximum elevation of the TSF will be 55 m and the rate of rise will be less than 2 m per year.

17.4 Metallurgical Recovery

The Loulo processing plant uses a CIL gold extraction process. The current throughput design capacity is 4.8 Mtpa as per 2017 performance (Table17-2). Yalea process plant recovery for the remaining LOM has been estimated at 84.5% based on historical and current testwork; similarly, Gara is estimated as 94.2%. Yalea recovery is impacted by the presence of arsenic and copper. Arsenic and copper impurities also increase cyanide and oxygen consumption. Gounkoto 2016 Super Pit sampling achieved a recovery of 92.5%. Gold recovery is maintained above 90% by blending the various ore sources (Yalea / Gara/ Gounkoto) to control the copper and arsenic content within the mill feed.

The current LOM has an average recovery 92.3%. The average gold recovery in 2017 was 92.7%, an improvement from 2016 (91.0%).

17.5 Processing Costs

Operating Costs (OPEX)

The open pit and underground LOM and operating cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

Table17-3 presents the 2016 and 2017 actual breakdown of costs for processing including plant engineering for Loulo-Gounkoto.

18^th September 2018 Page 273

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table17-3 Actual Process and Plant Engineering Operating Costs for 2016 and 2017

Cost	Units		2016 Actual		2017 Actual
Fixed Cost
Consultants		$ ‘000		166		153
Contractors - Assays		$ ‘000		742		798
Contractors - Other		$ ‘000		1,217		1,358
Contractors - Oxygen		$ ‘000		2,312		2,680
Equipment Hire		$ ‘000		2,537		2,523
General Costs		$ ‘000		1,142		1,520
Gold Refining		$ ‘000		1,564		1,646
Labour		$ ‘000		2,831		3,653
Stores - Electrical & Mechanical		$ ‘000		1,666		1,626
Stores - Other		$ ‘000		5,536		5,366
Total Fixed Cost		$ ‘000		19,712		21,323
Tonnes Processed		1000		4,875		4,918
Total Fixed Cost		$/t		4.04		4.34
Variable Cost
Power		$/t		5.49		6.21
Reagents - Cyanide		$/t		1.63		1.26
Reagents - Lime		$/t		0.77		0.62
Good Issues - Caustic Soda		$/t		0.11		0.15
Good Issues - Activated Carbon		$/t		0.16		0.13
Reagents - Other		$/t		0.33		0.34
Stores - Grinding Media		$/t		1.93		1.92
Stores - Screens & Panels		$/t		0.04		0.04
Total Variable Costs		$/t		10.46		10.67
Total Process Cost		$/t		14.50		15.01
Plant Engineering		$/t		2.19		2.20
Combined Plant & Engineering		$/t		16.69		17.21

LOM processing costs have been budgeted at $18.22/t for Loulo and $17.03/t for Gounkoto, which includes plant engineering costs. The actual costs for Loulo-Gounkoto for 2017 were $17.21/t, which is in line with those expressed for the LOM.

18^th September 2018 Page 274

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18 Project Infrastructure

Loulo-Gounkoto complex comprises two distinct mining areas, Loulo and Gounkoto, with the central processing plant and administration complex located at the Loulo site.

Current mining is centred on the Gounkoto open pit and two underground mines at Yalea and Gara.

The current access route to supply the site is by road from the port of Dakar in Senegal to the site. There is also access to Loulo Mine by road from Bamako or by charter flights from Bamako to the airstrip situated near the Gara deposit.

Figure18-1 denotes the relative positions of the major infrastructure items in the immediate vicinity of Loulo. Note that Gounkoto is not shown here as it lies approximately 32 km from Loulo in a Southerly direction. These infrastructure items will be described in greater detail subsequently.

Figure 18-1 Loulo Major Infrastructure Locations

18^th September 2018 Page 275

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18.1 Mine Roads

The Dakar to Bamako Millennium highway crosses the Loulo-Gounkoto haul road, approximately 6 km north of the Gounkoto pit. This highway serves as the primary access point for the mine and provide excellent road transport links with the rest of the country as well as to Senegal for which the border is within 3 km of the mine.

The local road infrastructure was developed initially during the geological drilling programmes and upgraded during the construction of the mine. Internal roads provide access to various infrastructure areas, Explosives Storage, Land Fill Site, Mine Villages (Senior and Junior), Central Mine Offices, general mining operations areas, new exploration areas, various water boreholes, and overhead line routes.

The road leading to the TSF is a public road, which runs past the nearby village of Djidian Kenieba. An upgrade exercise was undertaken whereby the roads linking the Mine village to the central office complex via a ring road ultimately culminating at the plant area has been tarred. All other roads are constructed by layered rock/gravel/laterite varying in specification according to traffic expectations.

18.2 Supply Chain

From the inception of Loulo-Gounkoto, the supply chain partner,CSTT-AO, with subsidiary companies namely Afrilog and Multilog, have been intricately involved in the procurement, warehousing, freight, and all other logistical aspects.

The port of Dakar is used for goods destined to Loulo. The customs process and port authorities are efficient and port costs are in line with global market. Dakar Customs and border Customs operate on afive-day week basis.

Should containers be shipped to Dakar, the majority are trans-shipped in Las Palmas or Algeciras and shuttle shipments between these trans-shipment ports to Dakar are frequent. The port is also well equipped to handle heavier materials and equipment which are frequently received for Loulo-Gounkoto.

Border crossing clearance of goods is typically seamless and efficient between Dakar and Mali customs. This relies heavily on the long-standing partnerships bothCSTT-AO and Randgold have developed over the past decades in West Africa.

As regards trucking and road freight,CSTT-AO is equipped with adequate trucks, lowbed trucks and lifting equipment capable of lifting loads of up to 100 tons (available in Dakar).

CSTT-AO has the following fleet capabilities:

● 90 tractors

● 90 flatbed and semi -trailers

18^th September 2018 Page 276

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

● One low bed of 110 tons capacity

● One low bed of 300 tons capacity

Where possible load heights are restricted to less than 485 cm including the vehicle height, to allow easy passage from the port to site, with no disruption to power lines or having to make use of bridge bypass routes.

CSTT-AO is equipped and licensed to transport all hazardous materials including being specifically cyanide compliant.

Clearing of freight at Dakar port takes on average eight to twelve days and the road transport from Dakar to the Project site is estimated at two to three days.

CSTT-AO has been developing activities at Kaolack port since November 2016. This is beneficial as Kaolack is 200 km closer to the Malian border than Dakar port.

The maximum draft allowed is 4.3 meters, (2,400 t of break bulk vessel).

To date, quick lime, hydrated lime, grinding media and slag cement have all been imported in break bulk and sent to the mining companies located in the vicinity of the South Eastern region of Senegal and Mali.

Clearing of goods is typically around 48 hours.

The costs associated with 20’ and 40’ containers for bothsea-freight and inland transport (Dakar to Loulo mine site) are calculated on a cost plus basis. This is a fully transparent exercise with shipping/freight invoices being sent through for verification.

18.3 Surface Water Management

There is an occasional torrential rainfall threat to the operations with the risk of flash flooding. Drainage and diversion ditches are in place around the pits and operational facilities.

There are river diversions at all the old and current pits. A trench designed for a 1 in100-year event has been constructed at Gara. At Yalea, the river runs to the west of the pit and a diversion channel has been dug on the east side to prevent water from creeks entering the pit. The ground is sloped away from the pit wall to further reduce the risk of flooding. The diversions remain in place to protect the underground workings.

Gounkoto is bounded on the west by the Falémé River and a 1.8 km berm and diversion channel have been constructed. On the east side of the pit a river diversion has been necessary to drain water around the pit to the north and then into the Falémé.

The wet season in Mali has torrential rain under which mining can be impossible. Care is taken with mine scheduling to take time into account for digging sumps, as well as providing more than one working area to allow for flexible mining if required. Ore stockpiled at the Gounkoto ROM pad

18^th September 2018 Page 277

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

and thelow-grade stockpile at Gounkoto can provide plant feed in the event of mine production having to be stopped due to weather conditions.

For the Gounkoto open pit, a high capacity pumping system of large high head diesel pumps has been installed. Pumping from pit bottom sumps to booster pumps delivers captured water to treatment ponds prior to releasing to river flow. During the wet season, pumping volumes average 5,000 m3/day from the southern sump and 4,000 m3/day from the northern sump.

18.4 Water Supply

The Falémé River is the natural border between Mali and Senegal and provides the majority of the Loulo water requirements.

A main pumping station with two transfer pumps is situated in the river basin upstream of a weir, which maintains a 1 m raise to ensure water availability during the dry season. Water is pumped to the Raw Water Dam at Loulo from where water is delivered to the process plant. There are no restrictions on the amount of water which can be abstracted from the river.

A water treatment plant, fed from the river water supply, supplies all the potable water needs of the operation. A borehole pump is situated near the offices to provideback-up potable water.

There is a very low requirement for water at Gounkoto, other than crushing of the ROM material before loading and dust suppression. Rainfall is regular, and the pit dewatering ponds provide enough water for dust suppression purposes. A small onsite dam provides water storage.

18.5 Tailings Facilities

The tailings storage facility (TSF) for Loulo is 8 km from the plant in an area with a number of natural ridges. It has been designed to maintain a minimum freeboard of 1.5 m to provide sufficient storage to contain a 1 in 50 year rainfall event over a72-hour period. The TSF was designed by Knight Piésold and is operated by Fraser Alexander.

The operation is based on a 42 paddock system, using two at any one time. Each paddock is 100 m in length and 35 m wide. Each lift is 200 mm and construction of the walls is completed by hand using deposited material, whilst there is anopen-end deposition. Each paddock is used for 36 to 48 hours.

It takes approximately one and a half months to complete an overall 200 mm lift of tailings in the TSF, which gives an annual rise of approximately 2.0 m (current rate of rise is 2.03 m/year). At approximately 10 m height, the paddock area was stepped in to leave an approximately 8 m wide bench.

The total area of the dam is currently 175.6 ha, and the current height is approximately 27 m. Since the start of operations approximately 33 Mt have been deposited into the TSF.

18^th September 2018 Page 278

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The current design LOM is 43 m. Should the LOM extend to 2037 the current design would accommodate a final height of up to 55 m Figure18-2 shows the starter wall built in the east side of the TSF.

Figure18-2 Starter Wall Built on East Side of TSF

Tailings are discharged from the relevant CIL tank, screened at the tails linear screen (0.8 mm by 0.8 mm) and report to the tails tank from where they are pumped to the either the Intermediate Paste Plant or tailings storage facility (TSF). The current split is approximately 40% / 60%. Total tailings production is approximately 12,000 tpd. The tails pump deliveries are equipped with a flow diversion valve, where the delivery lines of the duty and standby pipes combine into the tailings main pipe.

Two steel tailings lines from the plant and one HDPE return water pipe. The second steel line is used as aback-up line. All pipelines are of 450 mm diameter, lying above ground and inside a trench, along a vehicle access road between the TSF and the processing plant. The tailings are pumped by two 90 kW pumps with a total capacity of 620 m3/hr at an insitu density of 1.37 t/m3.

Return water is pumped from a floating barge with two electric pumps rather than a traditional penstock arrangement. It is estimated that the TSF retains approximately 25% of the water.

During the three months wet season, the TSF tends to absorb more water, which generally reduces during the remainder of the year. A marker system on top of the TSF is used to measure

18^th September 2018 Page 279

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

the pool size. Different colour markers are placed around the pool to monitor any increase in the size of the pool. Freeboard is measured weekly.

If required, excess water can be pumped into the detox pond by a 600 m3/hr pump. The detox pond has an overflow into theso-called “detox area” in the flood plain to the north of the TSF. There is a cyanide destruction installation at the detox pond and the cyanide content of the detox pond and the pool are sampled daily.

Monitoring piezometers surround the facility and are read manually at weekly intervals.

The TSF operation is audited regularly by Epoch from South Africa, who last visited in June 2017. No major areas of concern were noted other than an area of seepage on the western wall that if left unchecked could affect the Factor of Safety of the designed final height.

A further concern the mine is facing is the accumulation of water on the tailings dam, which potentially puts the stability of the dam at risk. Actions are in place to reduce the pool size and the necessary studies are being undertaken together with the submission of an Environmental Impact Assessment (EIA) to the Malian Government as part of an application to extend tailings dam.

18.6 Power Supply

All the power used on site is thermally generated, and fuel is imported from Senegal by road. Filtration and/or centrifugation are used to produce a ‘clean’ diesel which can be used in the mobile equipment and the power plant. At Loulo there are two main fuel farms; one near to the workshops and warehouse and the other at the power plant.

At the warehouse there are five 500 kl diesel tanks in a common bund and protected by a ring drencher system that has been recently upgraded. There are also five 50 kl tanks for clean diesel, four at the powerhouse and one for light vehicles.

The second major fuel farm is at the power plant for storage of HFO. All generators run on Heavy Fuel Oil (HFO) with diesel used as a starter fuel. The tank farm comprises:

● Three 2,000 kl tanks for HFO storage

● Two 230 kl HFO settling tanks

● One 253 kl day tank for filtered and settled HFO

● One 230 kl Light Fuel Oil (LFO) tank

Fire protection for the HFO tank farm and both power stations consists of a pump set with a main diesel pump with an electrical jockey pump to maintain pressure under normal circumstances. Water is supplied from a pump station dedicated to these areas.

At Gounkoto there are two 500 kl tanks and one 75 kl and one 45 kl clean diesel tanks. Detection and suppression are installed on the fuel farm; stationary monitors are surround the fuel farm and cooling deluge rings are in place on all storage tanks.

18^th September 2018 Page 280

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Current demand is approximately 42 MVA. The main power plant is located at Loulo and has an installed capacity of 64 MVA and is a mixture of medium speed and high-speed generators. The power station for the site is operated by Manutention Africaine Mali, part of the Caterpillar Africa Power Systems. A full maintenance contract is in place.

The plant has a mixture of medium speed and high-speed generators as follows:

● Fifteen CAT3512B-HB 1.2 MVA high-speed generators.

● Ten CM32 3.5 MVA medium-speed generators.

● Two CM32 5.5 MVA medium-speed generators.

The high-speed generators run on normal diesel. The medium speed generators are operated preferably on HFO180.

All CAT generators are installed within one power house building. Four of the CM 3.5 MVA generators are within another building adjacent to the original power house. The other six CM 3.5 MVA units are located in the new power house, together with the two CM32 5.5 MVA. There is sufficient space for a further two machines.

Power is distributed around the site at 11kV. A central medium voltage room is used as a switching room to distribute power across the mine through several feeders via dual 11 kV overhead lines. The CAT generators produce power at 400V, which is then stepped up to 11kV through individual transformers. The CM generators produce at 11kV. A central medium voltage room is used as a switching room to distribute power across the mine through several feeders via dual 11 kV overhead lines.

The site base load is normally generated by the CM generators and the CAT generators are used for peak loads to reduce the reliance on the transformers. Currently 60 to 70% of total generation capacity is needed to meet the site demand.

The Yalea and Gara underground mines share four 1.2 MVA rental units asback-up power for the critical underground facilities, such as pumps and ventilation.

The electrical demand at Gounkoto for the open pit mining operations is low, typically requiring around 1.2 MW. Two CAT 3560 1.2 MVA generators plus a further standby 0.8 MVA generator, together comprising 3.0 MVA are installed at Gounkoto and provide the power for the site. Construction of a new power plant at Gounkoto is well advanced, where initially it will have two 1.5 MVA generators but with space for a total of five generators when the future underground mine commences. Backup electrical power can be supplied to Gounkoto via a 33 kV overhead line from Loulo.

18^th September 2018 Page 281

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

18.7 Site Infrastructure

Operational Camp (Village)

The Loulo offices, warehouses and accommodation village are located to the east of the Gara open pit. Due to the remote nature of the operations there is a comprehensive provision of auxiliary facilities at Loulo. Living quarters include a mine village for the expatriate labour and senior staff.

A small clubhouse and mine recreation facility have been built on site which is the focal point of the mine living quarters. This structure is designed to blend into the environment. The recreation rooms are designed in a flexible manner so that they are used for all mine functions and training when required

Offices, Stores and Workshops

Extensive use has been made of old shipping containers for buildings either individually, or as part of a brick constructed building such as for the workshops. The mine site buildings are therefore a mixture of constructions. These comprise:

● Administration office buildings

● Warehouse and Stockyard

● Light vehicle workshop

● Laboratory

The warehouse at Loulo serves the Gounkoto Mine.

Both Gara and Yalea underground mines have small workshops near the surface portals where servicing of mobile mine equipment is carried out. A central workshop is used for more substantial maintenance. The mine workshops are used for preventative maintenance and specific component change outs while the Central workshop is responsible for major overhauls and large component change outs.

The Central Workshop is undergoing an expansion, increasing it to 22 bays. An overhead crane is installed. Sandvik has established a warehouseon-site where it holds inventory stock.

The mine workshop at Gounkoto is owned by SOMILO but is operated by the contractor, GMS. The workshop consists of six large bays and each bay can accommodate one CAT 777 truck, or up to four smaller mobile mining machines.

Medical and Emergency Response Facilities

There are three medical clinics available; one in Loulo with three doctors and nurses, one in Gounkoto, and one in the Junior Village. Each clinic has an ambulance.

18^th September 2018 Page 282

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

A number of local medical facilities have been identified, which could assist in case of multiple injuries. Planes and ambulances can be used for medical evacuations.

A Mine Rescue (Proto) Room is established and equipped with facilities for washing and storing the sets, a Drager pump for recharging the cylinders used in the BG4 sets and storing the kit of the individual members.

There are Proto teams in place to serve the underground operations at Loulo each comprising of ten persons. In addition, a turnout agreement has been concluded with the Byrnecut Proto Team at the neighbouring Tabakoto Mine which is owned by Endeavour.

A fully equipped fire truck is stationed at Loulo and is used at the airstrip when planes land. In addition, both at Loulo and Gounkoto mine water trucks are available to supply additional water and firefighting capability.

Airstrip

The airstrip is 1.5 km long and built from laterite. It has been approved by the National Directorate for Civil Aviation. The maximum size of aircraft that can use the airstrip is six tonnes per single wheel and twelve tonnes per double wheel. The airstrip is capable of taking aircraft with a typical capacity of 40 seats.

Randgold organises charter flights from Bamako twice per week, and these are usually20-seater Beechcraft 1900 planes.

18.8 Communication and Information Technology

Internet Communication

A VSAT has been installed at both Loulo and Gounkoto, and a network link connects the two sites. The IT/ Communications configuration networks Gounkoto into the Loulo network with a high-speed terrestrial network link with each site providing redundancy to the other.

The dual VSAT links provides redundancy in the event of either VSAT becoming temporarilynon-operational. The dual VSAT links also allow communication for their respective site, if the terrestrial link is severed.

The existing networking set up and wireless Internet communication is sufficient for the ongoing operation. Email is managed over the VSAT link with Gounkoto email directed to the Loulo mail server over a terrestrial link.

Mobile and Hand-Held Radios

A VHF radio system is installed to provide contact with all mobile equipment and all of the service vehicles. The VHF system has 8 channels with a computerdial-up channel selection. Use of five of the channels in the operation are allocated as follows:

18^th September 2018 Page 283

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

● Mining

● Metallurgy

● Grade Control

● Security

● Warehouse

The remaining channels are used for departments associated with the plant operation and maintenance. Hand held radios are used by roving operators and supervisors to ensure adequate communication.

18.9 Security

There is comprehensive security infrastructure at the site, with good access control commensurate with the operations at the site. The Security Manager reports directly to the onsite General Manager. A contractor, AMM, provides the security service. In addition, there are 18 local hunters being used to provide external security patrols.

The principal risks are ongoing loss of company equipment and stores by means of fraud and theft, mainly by employees and the ever-present risk of housebreaking and petty thefts in the residential village.

The Loulo mine property is partially fenced with 2.4 m high chain link fence topped by flat wrapped razor wire and a road runs along the entire perimeter. Site fencing is also provided at Gounkoto.

The plant area is fenced with the access controls at the main gate and to the more sensitive areas within the plant. An electronic access system for the plant is operation.

There is one main entrance to the Loulo site, where a Security Gatehouse has been erected and manned by three guards, on a 12 hour shift cycle. A similar arrangement exists at Gounkoto.

The Spares and Materials storage sites are fenced off by a 1.8 m fence topped by a 0.5 m barrier of flat-wrapped razor wire. The access gates are kept permanently locked and access is controlled by staff in an adjacent office.

18^th September 2018 Page 284

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

19 Market Studies and Contracts

19.1 Revenue, Tax and Royalty

Financial evaluation andcut-off grade calculation for the Loulo Ore Reserves has been based on a gold price of $1,000/oz. This is the same as was used for the previous (31st December 2017) Loulo and Gounkoto Ore Reserves estimates.

A royalty of 6% of the proceeds of gold sales is payable by Loulo-Gounkoto to the Mali government.

Loulo-Gounkoto pay income tax at a rate of 30% to the Mali government.

19.2 Markets

The gold market is highly liquid and benefits from terminal markets (London, New York, Tokyo, and Hong Kong) on almost a continuous basis. Gold prices were in general downward trend from 1980 to 2000 where it traded down to approximately $250/oz. Between 2000 and 2011 the market was on a general upward trend that moved spot prices to a peak of $1,900/oz. momentarily during 2012. 2013 saw a sharp correction in the upward trend, with the spot price dropping to $1,250/oz. Since 2014 the gold price, has fluctuated in range of $1,050/oz to $1,400/oz. Gold is currently (August 2018) trading at $1,200/oz.

Gold produced at the mine site is shipped from site, under secured conditions, to a refining company. Underpre-established contractual conditions, the refiner purchases the gold from the mine with the proceeds automatically credited to the mines’ bank account. The operation is unhedged.

19.3 Contracts

It is Randgold’s strategy to outsource open pit mining activities to contractors and, in all instances, the contract states that the mining operation can purchase the equipment at the end of the contract period at its depreciated price or should the contractor default at a predetermined pricing mechanism. Prior tostart-up all major mining contractors are requested to tender and the most appropriate tender is accepted thereby ensuring that the best competitive current pricing is achieved. Care is taken at the time of finalising contracts to ensure that the rise and fall formula is totally representative of thebuild-up of the quoted price per unit. At the time of award prices quoted are compared to benchmark prices of other owner miner operations.

The contract mining costs are dependent on when tenders are issued as the price of major equipment varies dependant on demand as well as the cost of finance. Rise and fall can be negatively affected by currency fluctuations.

18^th September 2018 Page 285

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The mine produces doré bars which are sent to an accredited gold refinery for refining. Refining prices are subject to fluctuations in the cost of transport as well as insurance costs. Other contracts that are put in place include assay facilities, oxygen supply, catering services, fuel supply, explosive supply, security.

18^th September 2018 Page 286

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

20 Environmental Studies, Permitting, And Social or Community Impact

20.1 Environmental Considerations

Loulo-Gounkoto is a complex of open pit and underground operations which is made up of two main sites, Loulo and Gounkoto, which are around 32 km apart and which have been operating since 2005 at Loulo and 2011 at Gounkoto. Gounkoto is operating under a separate Exploitation Permit from Loulo since 2012 with ore being toll treated at Loulo. Loulo mining activities consist of the Yalea and Gara underground mines, the main open pits being exhausted. Other satellite pits have included Baboto, P129, and Loulo 3. Gounkoto pit is being enlarged to form a ‘Super Pit’ and will be followed by underground operations. Ore from Gounkoto is crushed and stockpiled and trucked to Loulo plant, where it is further crushed and ground and passes through a gravity circuit and CIL plant and smelted to produce bullion. Tailings are pumped 8 km east to the Tailings Storage Facility (TSF); a portion of the tails are fed to the Yalea and Gara paste plants and pumped to the underground workings. The TSF is being raised at a rate of 2 m per year with the current height being 28 m. Its final elevation is designed to be 55 m. A TSF extension will be required to accommodate the current life of mine production and the Environmental and Social Impact Assessment (ESIA) associated with this will be developed in 2018. Other facilities at Gounkoto include accommodation, workshop, and offices, two diversion dams, and acut-off trench to divert flows from the east and prevent them entering the open pit. The Falémé River bounds the western edge of the site and lies along the border with Senegal. The waste rock dump (WRD)which lies between the pit and the river and has been designed with a system of underdrainage and seepage collection to avoid any potential discharges from the WRD into the Falémé River. There are other WRDs surrounding the pit. Loulo facilities include four fuel tank farms containing HFO and LFO.

Typically, oxide ore has been mined from the open pits while the underground operations are targeting sulphide ore.

Environmental Assessment and Permitting

The Environmental and Social department of the mine reports to the mine General Manager (GM) operational with a functional report to the Group Community and Environmental Officer, who provides technical support to ensure that the mine operation remains in compliance with Randgold standards, host country laws and regulations and when necessary to the IFC/World bank standards. Each environmental department develops its budget and programmes in consultation with the Group Community and Environmental Officer, who also undertakes regular audits and inspections to ensure the site is compliant.

An environmental management program is in place which covers both operations. The Loulo and Gounkoto operations are ISO 14001:2004 compliant and independently audited to continuously improve environmental management. External audits are also carried out for compliance with the International Cyanide Management Code (ICMC), although Randgold are not directly members

18^th September 2018 Page 287

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

of the Code the gaps identified in the last external audit, carried out in March 2017, relating to transportation, handling and storage, operations, decommissioning, worker safety, emergency response and community dialogue are being addressed.

A further ESIA was completed in 2014 and submitted for approval in 2015 for the underground (UG) operations at Gounkoto.

Subsequent to the approval of the 2014 underground ESIA, Randgold undertook a feasibility study to determine thetrade-off between underground mining and the expansion of the existing open pit. Based on economic conditions, it was determined that the expansion of the existing open pit (to be referred to as the proposed Super Pit) was the preferred alternative. An ESIA for the Super Pit was submitted and approved in 2016. This ESIA was compiled based on the 2014 underground ESIA.

All environmental permits are in place for the Loulo processing plant and both the Yalea and Gara underground mines, coveringinter aliaexplosives, the Baboto satellite pit and connecting road, the mine canteen, abstraction of water from the river and resettlement of the hamlet of Faraba. The key permit is an environmental clearance permit which is valid for five years, following an audit of the Environmental Management Plan (EMP), the five yearly EMP audit. Works can be halted if the operation is not in compliance with the audit findings and the agreed environmental clearance. The most recent permits for both Loulo and Gounkoto were granted on 21st March 2017.

Environmental Management and Monitoring

The Environmental and Social department of the mine reports to the mine general manager (GM) operational with a functional report to the Group Community and Environmental Officer, who provides technical support when and if needed to bring the mine into compliance with Randgold standards, host country laws and regulations and when necessary to the IFC/World bank standards. Each environmental department develops its budget and programmes in consultation with the Group Community and Environmental Officer, who also undertakes regular audits and inspections to ensure the site is compliant.

An environmental management program is in place which covers both operations The Loulo and Gounkoto operations are ISO 14001:2004 compliant and independently audited to continuously improve environmental management. External audits are also carried out for compliance with the International Cyanide Management Code (ICMC), although Randgold are not members of the Code Gaps identified in the last external audit, carried out in March 2017, relating to transportation, handling and storage, operations, decommissioning, worker safety, emergency response and community dialogue are being addressed.

A comprehensive water balance model is in place for Loulo (Yalea and Gara operations) which models flows, inputs and losses across the complex site, including the open pits and underground workings, plant, TSF, water management structures, offices, camp, and treatment facilities. The dynamic model also identifies river water use, dewatering water, discharges, gains, and losses identifies volumes of potential savings/recycling opportunities which would reduce abstraction

18^th September 2018 Page 288

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

from the river. A similar but simplified water balance is in place for Gounkoto which reflects the fact that no processing is taking place there. Abstraction, storage, use, and discharge are monitored, and records kept in line with the GRI reporting format used by the mine for annual reporting to the Malian government.

Routine environmental monitoring takes place across the sites, including air quality (dust deposition and PM10) water quality, noise, and carbon emissions. Other specialist monitoring includes volatile organic compounds (VOC) emissions from the fuel tank farms, aquatic diversity in the watercourses and water bodies around Loulo, including the Falémé river and the Gara dam, a bioaccumulation study of metals in fish tissues, and gaseous monitoring (CO, NO, NO2, SO2, H2S) of various stacks and exhausts (two gold room, two laboratories, scrapyard incinerator, air power plant and the Gara underground ventilation system). The fish studies noted above clearly showed the impacts of ASM activities on water quality and sediment quality with elevated levels of arsenic and mercury, which is used in the ASM processing Environmental incidents are noted in a register at each site, causes and responses identified and incidents closed out. A total of 25 incidents (24 minor and one moderate) were reported in 2017 at Loulo, of which 56% were related to spills of oil, chemicals, or waste water.

An annual Environmental and Social report is submitted to the Malian authorities, compiled in a format aligned with GRI (Global Reporting Initiative) requirements. These reports cover environmental incidents, biodiversity and soils, water quantity, quality of discharges, surface water and groundwater quality and trends, energy and carbon emissions, cyanide management, noise and dust management, water management, the Environmental and Social Management System and closure costs. The section on Community Development covers grievances, stakeholder dialogue, artisanal mining, education and training, community health, drinking water, food security and economic development.

Mine Rehabilitation and Closure

Rehabilitation is ongoing at various locations at Loulo and as at the end of 2017, approximately 124 ha of land has been rehabilitated. Loulo has a full functioning nursery that is utilised to grow several local species of tree that are used for rehabilitation. While many open pits are mined out and waste rock dumps are inactive, these are still in the process of being assessed for potential future underground operations so will not be rehabilitated until near the end of mine life.

Mine closure costs are updated each year, with increases or decreases in disturbed areas noted and costed. Closure costings assume that there should not be any water requiring treatment from the pits and underground workings; the calculations do not account for any value recovered from the sale of plant, steel or other material; infrastructure (i.e. brick buildings) at the mine camp and mine offices, as well as all roads, which will be left as part of the agribusiness project after the mine closes (see 20.2 Social Considerations); contractor laydown areas will be rehabilitated by the contractors as per the contractor’s agreement; rehabilitation of the administration and certain workshop areas assumes that topsoil will be spread over 50% of the area, 50% of the areas will be ripped and the entire area vegetated. Waste rock dumps (WRDs) will be shaped to angles less than 30° and 300 mm of topsoil will need to be utilised to rehabilitate the TSF on the top and side slopes of the facility, this topsoil will be placed on a 400 mm layer of saprolite to be utilised

18^th September 2018 Page 289

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

as a capping layer. All pit edges will be shaped to approximately 27° to a level of 20 m below ground surface and a berm constructed around the perimeter of the pits. These areas will be vegetated; post-closure ground and surface water monitoring is assumed to continue for five years, with sampling taking place on a quarterly basis, while vegetation monitoring and maintenance will take place for three years.

Total closure costs for the Loulo Permit are estimated at $29.4 M which includes a contingency of 10% to allow for areas which may have been undervalued or which have been overlooked.

The cost for rehabilitation and closure of the Gounkoto mine according to the Digby Wells calculation model is $9.7 M

Some local contractor rates foron-site earth moving machinery were used to calculate the 2017 closure costs.

The University of Bamako chemical department undertakes periodic sampling to determine water quality and paste pH of various water streams, waste rock, sediments, and tailings. Waters sampled include that from the open pits, underground workings, and process water. Waste rock from 13 dumps was sampled in 2016 and Acid Base Accounting and leachate tests carried out. Waste rock is made up of many different lithologies and contains metals such as arsenic, zinc, copper, chromium, and manganese. Despite considerable neutralising potential in most rock types, some lithologies have sulphur concentrations up to 12.5%, and thus less neutralising capacity. A database is being maintained and recommendations are made for the management of waste dumps and any limited ARD.

20.2 Social Considerations

The various ESIAs carried out have included assessments of impacts and benefits to local communities. A comparative socio-economic survey has been conducted to assess the change (positive and negative) in the surrounding local communities over an eighteen-year period (1997-2015). This showed that since 2002 there has been a significant influx of people into the area of the mine, especially in Djidian-Kenieba and Loulo villages. Population figures were gathered by the fieldwork teams of 1997, 2002, 2007 and 2015. In 1997, the population of Loulo, Sakola, and Djidian-Kenieba was 975 but by 2015 it had grown to 19 685. Influx peaked in 2007 but had decreased by 2015. Modernisation of the communities has increased along with trade and commerce. Educational and skill levels have also increased but access to land to grow crops or graze livestock, and access to natural resources have decreased with higher populations and large tracts of land occupied by the mine infrastructure. The quality of the housing stock has increased, along with transport infrastructure, access to water, access to healthcare and education.

Primary livelihoods are equally distributed between agriculture (40%), orpaillage (40%), followed by housekeeping (15%), salaried employment (11%) (mostly at Randgold), and small-scale trading (10%). Secondary livelihood activities were only practiced by 60% of households, with orpaillage still being the most common, followed by subsistence farming and small-scale trading.

18^th September 2018 Page 290

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

80% of surveyed households Indicated that they are primarily reliant on cash-generating activities. Most common sources of household income were derived from employment at Randgold (34%) followed by artisanal mining (25%) and farming (14%).

Employment and Procurement

The mine is a significant employer to members of the local communities. The underground mining operations contribute to extendedlife-of-mine, employment of local Malians and the growth of the Malian economy. Randgold’s policy is to promote nationals of the host country to manage the Project. Where locally qualified and experienced staff is not available, recruitment from elsewhere is undertaken, with the clear understanding that local personnel are given the training and experience required to allow them to replace the expatriates as soon as possible.

Randgold’s policy of promoting local employment also extends to its contractors. Figures for national employment in 2017 were 2,820 out of a total of 2,975 employer and contractor posts at Loulo and 1,175 out of 1,209 at Gounkoto. Unskilled labour is typically sourced from the local area while more skilled posts are filled by staff from elsewhere in Mali, including Bamako.

There is also a policy of promoting local procurement which also covers contractors. Where possible, goods and services are procured locally. This includes produce from the agribusiness which is purchased for use in the mine canteens.

Resettlement

Phases of economic displacement (loss of crops and trees) and the physical resettlement of some households has occurred during the life of the mine to date as it has expanded. Crops and trees are compensated at published rates; physical resettlement has involved providing affected households with a new home. The most recent resettlement was around Gounkoto and was completed in 2012. This affected 12 households and around 300 hectares of land, including 1,700 economic trees and two artisanal mining (orpaillage) sites. A resettlement audit was carried out in late 2015, in line with good international industry practice.

Stakeholder Engagement

Stakeholder engagement and dialogue is ongoing and the importance of this to the company is demonstrated by the fact that senior managers have Key Performance Indicators (KPIs) related to their involvement in dialogue with communities, which generally requires them to attend quarterly stakeholder meetings. In addition, union representatives attend strategy meetings and board meetings of the local operating company in order to provide information on current community and local employee issue and suggest targets for the company to work towards with respect of the local community.

A grievance mechanism is in place. No grievances were received in 2017. The last grievance was recorded in 2015. Through the stakeholder engagement process, concerns are raised by the community: In 2017, the concerns registered were the generation of dust in the villages and the employability of young people in the community, a recurrent theme of these meetings. The issue

18^th September 2018 Page 291

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

of dust generation has been mitigated through the application of molasses, which acts as a binder, on roads through the riverside villages.

To meet the employment needs of youth in the community, strategies have been identified through the implementation of economic development projects. These economic development projects bring young people to entrepreneurship that will allow them to become self-employed and to become not a mine employee but rather a partner.

Community Development

Randgold continually engages with local communities and focuses on potable water supplies, primary school education, health care education, investment in medical clinics and local economic development projects. Hygiene and sanitation committees have been established to increase awareness. Randgold and its partners have built several classrooms in primary schools in villages around the mine.

A program to improve the agricultural yield of the area has been undertaken, with tractors being donated to the community and an agribusiness training centre constructed to train locals in agricultural entrepreneurship. A community of vegetable growers has developed to supply the mine caterer with fresh produce. The German Corporation for International Cooperation recently joined the mine’s effort in providing one million euros and their technical assistance.

Artisanal and Small-Scale Mining (ASM)

There is a significant ongoing presence of artisanal miners (orpailleurs) operating within the Loulo Permit area, particularly along the haul road. The government of Mali, as part owners of the operations, provide security in the form of Gendarmes who are posted at several locations in guard posts. The Gendarmes intervene on request if ASM activities interfere with the operation of the mine. The Gendarmes have been successful in removing orpailleurs from the Baboto target area. Although this and other key exploration targets remain free of ASM activity, the risk of incursion into exploration or operations areas remains, because of the increasing number of people in the illegal activity. The mining industry in Mali has established a committee to deal with the issue at the highest level of the Government. Dedicated orpaillage corridors are still to be created for artisanal and small-scale mining. The World Bank has recently been involved in proposing solutions. In the meantime, the Mine is reinforcing its relationship with the community to manage the issue.

18^th September 2018 Page 292

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

21 Capital and Operating Costs

21.1 Capital Costs

Basis of Estimate

Loulo is anon-going combined open pit and underground mining operation with the necessary facilities, equipment, and manpower in place to produce gold.

The basis for the combined LOM plan is the Proved and Probable Ore Reserves estimate described in Section 15.

In the QP’s opinion, the open pit and underground LOM and cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

The majority of the capital cost estimates contained in this report are based on quantities generated from the open pit and underground development requirements and data provided by Randgold.

Capital expenditure over the remaining LOM is estimated to be $598 M, made up from the following allocation of costs. A breakdown of the expenditure is detailed in Table21-1. The capital schedule is given in Table21-2 and Table21-3.

Table21-1 LOM Capital Expenditure

Capital Expenditure	Cost ($ M)
Construction and Project Capital	1.4
Ongoing Capital	107
Underground Capital Development and Drilling	311
Pre-Production Capitalised	137
Exploration Capitalised	2.9
Rehabilitation/Mine Closure	39
Total	598

Construction and Projects

A small amount of $1.4 M is allocated for projects other than the Gounkoto underground mine.

Ongoing Capital

Ongoing capital of $107 M is allocated to cover the overhaul and replacement costs for mining equipment other capital items.

18^th September 2018 Page 293

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Underground Capital and Development

The development of the Gounkoto underground operation is forecast to cost $74 M. A further $237 M is planned to cover capital development at the Gara and Yalea operations. Capital development costs are based on a calculated average cost per metre for development including development of declines, decline stockpiles, ventilation drives, level access drives, and long hole ventilation raises.

Pre-Production Capital

Pre-production capital covers open pit waste stripping. This was estimated at $137 M for Gounkoto. Exploration s capitalised where a total of $2.9 M was estimated for Loulo based exploration.

Exploration Capitalised

A total of $2.9 M was estimated for Loulo based exploration.

18^th September 2018 Page 294

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table21-2 Loulo Capital Cost

Item	Units	Total	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029
Construction & Project Capital	$‘000	1,410	304	304	290	256	256	-	-	-	-	-	-	-
Ongoing Capital	$‘000	56,859	34,459	3,800	3,800	3,800	3,800	3,600	3,600	-	-	-	-	-
Underground Capital + Development	$‘000	236,671	45,789	52,306	41,646	34,530	33,936	19,418	9,046	-	-	-	-	-
Pre-Production Capitalised	$‘000	-	-	-	-	-	-	-	-	-	-	-	-	-
Exploration Capitalised	$‘000	2,927	2,927	-	-	-	-	-	-	-	-	-	-	-
Rehabilitation/Mine Closure	$‘000	29,400
Total Capital Expenditure	$‘000	327,267	83,479	56,410	45,736	38,586	37,992	23,018	12,646	-	-	-	-	-
Table21-3 Gounkoto Capital Cost
Item	Units	Total	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029
Construction & Project Capital	$‘000	-	-	-	-	-	-	-	-	-	-	-	-	-
Ongoing Capital	$‘000	50,213	10,035	7,248	15,835	9,797	7,298	-	-	-	-	-	-	-
Underground Capital + Development	$‘000	73,920	-	-	-	-	-	-	-	30,734	30,792	8,483	3,208	704
Pre-Production Capitalised	$‘000	136,907	5,825	-	65,167	44,426	21,488	-	-	-	-	-	-	-
Exploration Capitalised	$‘000	-	-	-	-	-	-	-	-	-	-	-	-	-
Rehabilitation/Mine Closure	$‘000	9,700
Total Capital Expenditure	$‘000	270,740	15,860	7,248	81,002	54,223	28,786	-	-	30,734	30,792	8,483	3,208	704

18^th September 2018 Page 295

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

21.2 Operating Costs

Basis of Estimate

Randgold maintains detailed operating cost records that provide an excellent basis for estimating future operating costs.

The basis for the combined LOM plan is the Proved and Probable Ore Reserves estimate described in Section 15.

In the QP’s opinion, the open pit and underground LOM and cost estimates have been completed in sufficient detail to be satisfied that economic extraction of the Proved and Probable Ore Reserves is justified.

Costs used for the open pit optimisations were derived from the Mining Contractor’s pricing of the open pit LOM schedule. Owners cost were also added. Labour costs for national employees were based on actual costs. Local labour laws regarding hours of work, etc. were also considered and overtime costs included.

During 2017 costs for processing and G&A were updated based on actuals adjusted for latest forward estimates, production profiles and manning levels.

Customs duties, taxes, charges and logistically costs are included.

LOM Operating Costs

The LOM combined open pit and underground operating cost for Loulo is estimated to be $68.90/t milled and for Gounkoto it is estimated to be $68.49/t. Unit costs used to estimate LOM operating costs are summarised Table21-4 in for Loulo and Gounkoto. Total LOM operating costs for Loulo and Gounkoto are summarised in Table21-5.

Table21-4 LOM Operating Unit Costs for Loulo and Gounkoto

Activity	Units	Loulo	Gounkoto
Open Pit Mining	$/t mined	2.69	2.64
Open Pit Mining	$/ore t mined	34.10	36.95
Underground Mining	$/t mined	45.65	57.99
Underground Mining	$/ore t mined	46.35	82.40
Stockpile Movement	$/t milled	-3.63	3.81
Processing	$/t milled	18.22	17.03
Trucking & Hauling	$/t milled	0.03	5.20
G&A	$/t milled	8.30	7.73
Mining Total	$/t milled	42.35	38.53
Total LOM Net OPEX	$/t milled	68.90	68.49

18^th September 2018 Page 296

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table21-5 LOM Operating Total Costs for Loulo and Gounkoto

Description	Loulo LOM Total Cost ($ ‘000)	Gounkoto LOM Total Cost ($ ‘000)
Open Pit Mining	183,304	596,433
Underground Mining	1,326,789	174,447
Processing	649,829	340,790
G&A	295,816	154,730
Trucking & Hauling	1,225	104,036
Total Operating Cost	2,456,962	1,370,435

18^th September 2018 Page 297

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

22 Economic Analysis

This section is not required as the property is currently in production, Randgold is a producing issuer, and there is no material expansion of current production. RPA has verified the economic viability of the Ore Reserves via cash flow modelling, using the inputs discussed in this report.

18^th September 2018 Page 298

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

23 Adjacent Properties

23.1 Tabakoto Gold Mine

Endeavour Mining Corporation (Endeavour) own the operating Tabakoto gold mine which is approximately 26 km SE of the Loulo mine. Tabakoto reported Measured and Indicated. Mineral Resources on 31st December 2017 of 73.9 Mt at 1.56 g/t Au containing 3.60 Moz of gold. An additional 7.4 Mt at 3.4 g/t Au for 0.8 Moz of Inferred Resources was also declared. A Probable and Proved Ore Reserve of 4.8 Mt at 3.36 g/t Ay for 0.52 Moz was reported. Mineral Resources are inclusive of Ore Reserves. These Mineral Resources and Ore Reserves consist of both open pit and underground material.

The Tabakoto mine, like the Loulo mine, lays within the Kofi formation on the Kedougou-Kenieba erosional inlier. The mine began production in 2006, although problems withstart-up problems resulted in the mine closing in 2007. The minere-commenced operations in 2009 and has been operational since.

23.2 Kofi Exploration Project

Endeavour own, through Mines de Kofi SA, a local subsidiary, own a series exploration licence that lie directly north of the Loulo Permit. A technical report on their Tabakoto gold Mine dated March 2016 outlined that no Mineral Resource or Ore Reserve had been declared on their exploration permits. During 2017, Endeavour entered into an agreement with Randgold to purchase the Baboto North deposit which is adjacent to Endeavour’s Kofi C deposit. Endeavour expects to initiate mining activities at Baboto North in late 2018. No additional public information is available.

18^th September 2018 Page 299

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

24 Other Relevant Data and Information

No additional information or explanation is necessary to make this Technical Report understandable and not misleading.

24.1 Country Risk

The following is taken from the Annual Report on Form20-F 2017 submitted to the US Securities and Exchange Commission.

Randgold are subject to risks associated with operating gold mines in Mali. In 2017, gold produced in Mali represented approximately 58% of Randgold’s Consolidated group gold production, including joint ventures.

On March 21, 2012, Mali was subject to an attempted coup d’état that resulted in the suspension of the constitution, the partial closing of the borders and the general disruption of business activities in the country. The supply of consumables to our mines in Mali was temporarily interrupted as a result of the political situation. The borders were reopened shortly after these events and an interim government was installed within a month.

In January 2013, following military conflicts with terrorist insurgents, the Malian State requested the assistance of the French Government to assist the Malian army to repel the insurgents who had been occupying parts of the north of the country and beginning to move towards the southern part of the country. During 2013, French and other foreign troops occupied the northern part of the country to assist the Malian State in maintaining control of this region and presidential and parliamentary elections took place during the middle of 2013.

During 2015, a number of attacks by insurgents took place. Despite a peace agreement reached in June 2015 between the Malian government and secular armed groups, the growing presence of armed groups in northern and central Mali and bouts of violence have continued.

In July 2016, Mali extended the country’s state of emergency after a series of deadly attacks.

During 2017, Mali experienced a number of attacks by insurgents, including an attack on peacekeeping troops in the north of the country. In April 2017 and October 2017, Mali extended the country’s state of emergency as there continues to be a threat to security in certain areas of the country. Although we have continued to produce and sell gold throughout this period, there can be no assurance that the political or security situation will not disrupt our ability to continue gold production, or our ability to sell and ship our gold from our mines in Mali. Furthermore, there can be no assurance that the political and security situation in Mali will not have a material adverse effect on our operations and financial condition.

Goods are supplied to Randgold’s operations in Mali primarily by road through Senegal and Côte d’Ivoire, which at times have been disrupted by geopolitical issues. Any present or future policy

18^th September 2018 Page 300

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

changes in the countries in which Randgold operates, or through which Randgold are supplied, may in some way have a significant effect on Randgold’s operations and interests.

18^th September 2018 Page 301

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

25 Interpretation and Conclusions

Total production at Loulo-Gounkoto in 2017 was 4,918 kt at a head grade of 5.0 g/t Au producing 740 koz of gold (92.7% recovery). Total production from the mine as of 31st December 2017 totals 45,598 kt at a head grade of 4.2 g/t Au producing 5,637 koz (90.2% recovery).

25.1 Geology and Mineral Resources

The Loulo and Gounkoto has documented standard procedures for the drilling, logging, and sampling processes which meet industry standards. The geological and mineralisation modelling at Loulo and Gounkoto is based on visibly identifiable geological contacts, which ensure a geologically robust interpretation.

The Loulo and Gounkoto has a quality control program in place to ensure the accuracy and precision of the assay results from the analytical laboratory. Checks conducted on the quality control database indicated that the results are of acceptable precision and accuracy. In the QP’s opinion, the sample selection, preparation, and analysis are suitable for use within a Mineral Resource estimate.

Geological models and subsequent Mineral Resource estimations have evolved and improved with each successive model update. Significant grade control drill programs and mapping of exposures within mine developments have been completed to increase the confidence in the Mineral Resource and Ore Reserve estimates. During the last three years, significant resource extension drilling has been undertaken at both Yalea and Gara. This has led to the addition of new extensions Yalea Far South, Gara Far South, and Gara Far South Extended, making a major impact on replenishing 2016 and 2017 depletion.

In the QP’s opinion, both the Loulo Mineral Resources and Gounkoto Mineral Resources top cutting, domaining and estimation approach are appropriate and use industry accepted methods. The QP’s consider the Mineral Resources at Loulo and Gounkoto are appropriately estimated and classified.

The QP is not aware of any environmental, permitting, legal, title, socioeconomic, marketing, fiscal, metallurgical, or other relevant factors, which could materially affect the Mineral Resource estimate.

Exploration at Loulo-Gounkoto is focussed on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit extensions of the current Mineral Resources for high-grade mineralisation based on the structural model. During 2017, an updated tectonostratigraphy of the Kofi Series rocks at Loulo-Gounkoto has been developed, to improve the model of regional geologic architecture and understanding of key controls of gold deposit formation, which has subsequently driven are-assessment of greenfields exploration targets. The current exploration concept has been proven to be effective, with both

18^th September 2018 Page 302

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

the discovery of Gounkoto and the successful replenishment of depleted Mineral Resources and Ore Reserves at both mines.

25.2 Mining and Ore Reserves

The open pit operations at Loulo-Gounkoto consists of multiple open pits. The open pits are operated by a mining contractor and adown-the-hole blasting service is provided by an appropriate blasting contractor. Opportunities exist to upgrade and convert Inferred Mineral Resource within the current pit shells to Ore Reserves with drilling. The open pit operations provide a source of flexibility to supplement the plant feed when required. The Gounkoto open pit mining will be completed in 2024 with the Faraba Pit starting in 2022 and ending in 2024.

The Loulo underground mines are mature operations designed to extract the Yalea and Gara orebodies. The Loulo underground mines use three primary mining methods consisting of long hole transverse open stoping; long hole longitudinal retreat open stoping; and stoping under rock fill (SURF). The Yalea LOM extends to 2028 and Gara to 2032. Yalea and Gara sustain production rates of approximately 1.45 Mtpa and Gara 1.25 Mtpa, respectively.

Gounkoto underground is atPre-feasibility level. The proposed mining method for Gounkoto underground consists of longhole bench stoping with backfill. Due to the relatively small reserve tonnage of Gounkoto underground, all attempts to reduce the capital outlay have been adopted. The MZ3 mineralisation located on the footwall of the Fault Gouge fault has been removed from the schedule due to the poor ground conditions modelled in the area. Construction of the Gounkoto underground operation is currently scheduled to start construction at the end of the open pit mine life in 2024, achieve first production in 2025, and continue until 2030. This is currently expected to be reviewed during 2018. The Gounkoto underground operation will commence in 2025 and will continue until 2030.

Randgold has significant experience in other mining operations within West Africa and the production rates, modification factors, and costs are benchmarked against other West African operations.

The current Ore Reserves for Loulo-Gounkoto support a total mine life of 15 years, mining a total of 52.2 Mt of ore at 4.73 g/t Au until the year 2032. During this15-year period the total plant ore feed tonnage will amount to 56 Mt, including current stockpiles, at an average feed grade of 4.55 g/t Au resulting in 7.5 Moz recovered at an average processing recovery of 92%.

The QP considers the parameters used in the Mineral Resource to Ore Reserve conversion process to be appropriate.

The QP is not aware of any environmental, legal, title, socioeconomic, marketing, mining, metallurgical, fiscal, infrastructure, permitting, that could materially affect the Ore Reserve estimate.

18^th September 2018 Page 303

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

25.3 Processing

Based upon both metallurgical testwork data and actual operational evidence, the QP is satisfied that Loulo-Gounkoto is able to maintain production, gold recovery and reagent consumptions as forecasted.

The Loulo-Gounkoto has demonstrated successful operation both in terms of processing throughput and in particular with gold recovery.

Yalea recovery is impacted by the presence of arsenic and copper. Consequently, arsenic and copper estimations are completed as part of the Mineral Resource update in order to identify potentially low recovery areas. Gold recovery is maintained above 90% by blending the various ore sources (Yalea / Gara/ Gounkoto) to control copper and arsenic grades in the mill feed.

The current LOM has an average recovery life of mine of 92.3%. The average gold recovery in 2017 was 92.7%, an improvement from 2016 (Figure1-2).

The current LOM has an average recovery life of mine of 92%. Above 90% recovery is achieved by blending the various ore sources (Yalea / Gara/ Gounkoto) to control copper and arsenic grades in the mill feed.

The QP consider the modelled recoveries for all ore sources and the process plant and engineering unit costs applied to the Mineral Resource and Ore Reserve process to be acceptable.

25.4 Environment and Social

Loulo-Gounkoto has a mature environmental and social management plan and an accredited ISO14001 Environmental Management System in place which addresses current operational needs and can readily be adapted to meet future activities. Mine closure costs are reviewed and revised annually in line with good industry practice.

All permits are in place, and an annual Environmental and Social report compiled in a format aligned with GRI requirements is submitted to the Malian authorities.

Stakeholder engagement is ongoing, and all senior management are involved in regular meetings with the community. The mine prioritises local employment and in 2017 achieved 96% Malian employment across Randgold and contractor workforces.

Randgold continues to invest in community development initiatives, focussing on potable water supplies, primary school education, health care education, investment in medical clinics and local economic development projects, and livelihood projects, such as the program to improve the agricultural yield of the area. The mine is a significant employer to members of the local communities. The mining operations contribute to extendedlife-of-mine, employment of local Malians and the growth of the Malian economy. Randgold’s policy is to promote nationals to manage the Project.

18^th September 2018 Page 304

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

The ongoing presence of artisanal miners (orpailleurs) operating within the Loulo Permit area, has the potential to cause unrest. The government of Mali provides security in the form of Gendarmes who intervene on request if ASM activities interfere with the operation of the mine. The risk of incursion into exploration or operations areas remains, because of the increasing number of people in the largely illegal activity. Dedicated orpaillage corridors have been proposed by the Mine but are still to be created for artisanal and small-scale mining. In the meantime, the Mine is reinforcing its relationships with the community to deal with the issue.

The QP considers the extent of all environmental liabilities, to which the property is subject, to have been appropriately met.

25.5 Ownership

All Permit fees, surface rights fees, and taxes relating to Loulo-Gounkoto exploration and mining rights have been paid to date and reporting requirements have been met.

At the time of compiling this report, the QP is not aware of any risks that could result in the loss of ownership of the deposits or loss of the Permits, in part or in whole.

25.6 Infrastructure

As a result of the long-established open pit mining operation at the Project, substantial infrastructure exists at Loulo-Gounkoto to support theon-going mining and processing operations.

In comparison to much of the country, road access for staff and materials is excellent via the recently constructed Millennium Highway that crosses the Loulo to Gounkoto haul road approximately 6 km north of Gounkoto.

There is sufficient power supply capacity available from theon-site light and heavy fuel oil generators to meet the power demands of the operations.

An adequate water supply is available for the operation, sourced from the Gara and Falémé rivers which run through the Project site.

25.7 Risks

Randgold has undertaken analysis of the Project risks. summarises the Project risks and the QPs assessment of the risk degrees and consequences, as well as ongoing/required mitigation measures. The QPs, however, note that the degree of risk refers to our subjective assessment as to how the identified risk could affect the achievement of the Project objectives.

Loulo underground has been in production for 11 years and is a mature operation. Gounkoto open pit has been in production for over seven years and is a mature operation. Apre-feasibility study has been completed on Gounkoto Underground.

18^th September 2018 Page 305

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

In the QPs opinion, there are no significant risks and uncertainties that could reasonably be expected to affect the reliability or confidence in the exploration information, Mineral Resource or Ore Reserve estimates.

Risk Analysis Definitions

The following definitions have been employed by the QP’s in assigning risk factors to the various aspects and components of the Project:

● Low– Risks that are considered to be average or typical for a deposit of this nature and could have a relatively insignificant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

● Minor– Risks that have a measurable impact on the quality of the estimate but not sufficient to have a significant impact on the economics. These generally can be mitigated by normal management processes combined with minor cost adjustments or schedule allowances.

● Moderate– Risks that are considered to be average or typical for a deposit of this nature but could have a more significant impact on the economics. These risks are generally recognisable and, through good planning and technical practices, can be minimised so that the impact on the deposit or its economics is manageable.

● Major– Risks that have a definite, significant, and measurable impact on the economics. This may include basic errors or substandard quality in the basis of estimate studies or project definition. These risks can be mitigated through further study and expenditure that may be significant. Included in this category may be environmental/socialnon- compliance, particularly in regard to Equator Principles and IFC Performance Standards.

● High– Risks that are largely uncontrollable, unpredictable, unusual, or are considered not to be typical for a deposit of a particular type. Good technical practices and quality planning are no guarantee of successful exploitation. These risks can have a major impact on the economics of the deposit including significant disruption of schedule, significant cost increases, and degradation of physical performance. These risks cannot likely be mitigated through further study or expenditure.

In addition to assigning risk factors, the QPs provided opinion on the probability of the risk occurring during the LOM. The following definitions have been employed by the QPs in assigning probability of the risk occurring:

● Rare– The risk is very unlikely to occur during the Project life.

● Unlikely– The risk is more likely not to occur than occur during the Project life.

● Possible– There is an increased probability that the risk will occur during the Project life.

● Likely– The risk is likely to occur during the Project life.

● Almost Certain– The risk is expected to occur during the Project life.

18^th September 2018 Page 306

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Risk Analysis Table

Randgold has undertaken analysis of the Project risks. Table25-1 and Table25-2 summarise the Project risks and the QPs assessment of the risk degrees and consequences, as well as ongoing/required mitigation measures. The QPs, however, note that the degree of risk refers to our subjective assessment as to how the identified risk could affect the achievement of the Project objectives.

Table25-1 Loulo Risk Rating

Issue	Likelihood	Consequence           Rating	Risk Rating	Mitigation
Geology and Mineral Resources – Confidence in Mineral Resource Models	Unlikely	Minor	Low	Additional scheduled infill drilling. Resource model updated on a regular basis using production reconciliation results.
Mining and Ore Reserves – Open Pit Slope Stability	Unlikely	Minor	Minor	Continuedin-pit monitoring, geotechnical drilling, instrumentation, and continued updating of geotechnical models.
Mining and Ore Reserves – Underground Recovery and Dilution	Possible	Moderate	Low	Extensive production drilling for stope design, including geotechnical features. Predictive modelling of dilution.
Processing – Process Water Management with Mine in Water-Positive Mode	Possible	Moderate	Moderate	Water balance investigations Processing to reduce high usage of fresh water I-plant water treatment has reduced deleterious elements
Processing – Baboto oxide effect on paste fill	Unlikely	Moderate	Low	Several test trials campaigns already successfully completed, which has identified the maximum content of Baboto ore that can accompany a plant feed. Plant trials have demonstrated that blending of oxides, provided that threshold contents are not exceeded, will not compromise the paste plant operation.
Environmental – TSF Failure	Possible	Moderate	Moderate	Temporary holding of supernatant in redundant pits. Expansion of TSF. Plant Water balance being addressed. Pool management and removal of excess water off the dam

18^th September 2018 Page 307

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Issue	Likelihood	Consequence           Rating	Risk Rating	Mitigation
Environmental – Groundwater contamination (As)	Possible	Moderate	Moderate	Manage As levels of underground and TSF discharge through appropriate dilution techniques. Recycle contaminated water Continuing monitoring and external or third-party audits.
Social – Social License to Operate	Possible	Moderate	Low	Regular community and union engagement by company social and sustainability representatives.
Social – Artisanal Miners	Likely	Moderate	Moderate	Engagement with Malian Government.to agree an ASM strategy. Gendarmes securing Permit area.
Country & Political – Security – Governmental	Possible	Major	Low	Dedicated government liaison team in Bamako. Government participation/ownership.
Capital and Operating Costs	Possible	Minor	Low	Continue to track actual costs and LOM forecast costs, including considerations for inflation and foreign exchange. Owner/Operator for underground mining.
Fiscal Stability	Possible	Moderate	Moderate	Dedicated government liaison team in Bamako. Government participation/ownership

18^th September 2018 Page 308

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Table25-2 Gounkoto Risk Rating

Issue	Likelihood	Consequence           Rating	Risk Rating	Mitigation
Geology and Mineral Resources – Confidence in Mineral Resource Models	Unlikely	Minor	Low	Additional scheduled infill drilling. Resource model updated on a regular basis using production reconciliation results.
Mining and Ore Reserves – Open Pit Slope Stability	Unlikely	Major	Moderate	Improved blast hole control. Continuedin-pit monitoring.
Mining and Ore Reserves – Underground Recovery and Dilution	Possible	Moderate	Moderate	Further studies as part of the Feasibility Study.
Environmental – Air Quality (Dust)	Likely	Minor	Moderate	Dedicated dust suppression on haulage roads.
Social – Social License to Operate	Possible	Moderate	Low	Regular community and union engagement by company social and sustainability representatives.
Social – Artisanal Miners	Likely	Moderate	Moderate	Engagement with Malian Government.to agree an ASM strategy. Gendarmes securing Permit area.
Country & Political – Security – Governmental	Possible	Major	Low	Dedicated government liaison team in Bamako. Government participation/ownership.
Capital and Operating Costs	Possible	Minor	Low	Continue to track actual costs and LOM forecast costs, including considerations for inflation and foreign exchange.
Fiscal Stability	Possible	Moderate	Moderate	Dedicated government liaison team in Bamako. Government participation/ownership

18^th September 2018 Page 309

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

26 Recommendations

The QPs make the following recommendations:

● To extend the Loulo-Gounkoto life and replace depleted reserves, the current exploration strategy of targeting extensions of existing brownfields targets should be continued.

● A digital logging and data capture system should be implemented in the near future to minimise manual data capture.

● To reduce the complexity ofsub-domaining and improve the quality of local estimation, dynamic anisotropy should be tested and applied to folded bodies in future Mineral Resource updates.

● The application of a variablecut-off grade at Gara should be reviewed to avoid to the inclusion of large tonnages of lower grade material from the deeper levels of the mine impacting the profitability of the Gara operation.

● The Gara DRS model should be compared to production actuals, in order to establish if it could be directly applied to the Ore Reserve.

● To minimise the possibility of increased mining induced stresses affecting pillar and stope stability, a geotechnical review (including numerical modelling if appropriate) of the current Loulo underground mine plan and stoping sequence should be completed.

● The costs being used to calculate thecut-off grade for Gounkoto underground needs to be updated with the current underground costs and thecut-off grade applied to the next Mineral Resource and Ore Reserve estimates.

● It is recommended that a QP development programme be implemented at the Loulo- Gounkoto for resource geologists, mining engineers, and metallurgists.

● Loulo-Gounkoto is reliant on its own power generation through thermal energy sources. It is recommended that trials of alternative energy solutions continue to be sought as a potential reduction in power costs could have a material impact on the operating costs and mine life.

● An ASM strategy should be sought with the Malian government, and dedicated ASM corridors defined and managed.

18^th September 2018 Page 310

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

27 References

1008_Randgold_Loulo_22a_Yalea_Oxide-Tran_Options_140906_v04, Kenmore Mine Consulting, November 2013-December 2014.

1008_Randgold_Loulo_22b_Yalea_South_under dyke_mining_140806_v03, Kenmore Mine Consulting, November 2013- December 2014.

Air Quality Assessment Loulo 2016.

Air Quality Assessment Loulo 2017.

Annual Reports from 2015, 2016 and 2017.

Audit of Yale and Gara Mineral Resources, Loulo Gold Mine, Mali (DRAFT1), Code- RRS 21304, QG (Australia) Ltd, November 2014.

Competent Persons Report Mineral Resources, Loulo Gold Mines, Mali. Compiled by Simon Bottoms, Group Resource Geologist, Randgold Resources Limited, 31st December 2017.

Environmental Impact Assessment & Environmental Management Program Report, Loulo Gold Mine (Mali), Underground Mining Operations, Digby Wells & Associates, South Africa, October 2008.

Environmental Permit Gounkoto 2017.

Environmental Permit Loulo 2017.

Gara Conveyor and Vehicle Declines Stress Modelling, Chris Nyoni, Randgold Resources Ltd, March 2015.

Gara Crusher Chamber Stress Modelling, Chris Nyoni, Loulo, Randgold Resources Ltd, February 2015.

Geotechnical report on the Stability and mine design of Yalea underground, Middindi Consulting Pty Ltd, Ruimsig South Africa, ReportRDG-12Lou, March 2011.

Golder 2011, Loulo Gold Mine – Paste Backfill Scoping Study, Report11-1900-1005 (4000), Golder Associates, Sudbury, Ontario, Canada, October 2011.

Gounkoto Annual Closure Cost Assessment 2017.

Gounkoto Closure Liability 2017.

Gounkoto Incident Register 2017.

Gounkoto Resettlement Framework Oct 2010.

18^th September 2018 Page 311

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Gounkoto, 2014. QA/QC Report – Gounkoto QA/QC Period 1st October 2013 to 30th June 2014. Internal SMG report (author not listed).

JORC code 2012 Edition, Australasian Code for Reporting of Exploration Results, Mineral Resources and Ore Reserves, Joint Ore Reserve Committee of The Australian Institute of Mining and Metallurgy, Australian Institute of Geoscientists and Mineral Council of Australia (JORC), 2012.

Kenmore 2013, PowerPoint presentation- 1008_Randgold_Loulo UG review Sched optimisation_stage 1_131114_v02, Kenmore Mine Consulting, November 2013.

Kleynhans, J. and Abdoulaye, N. 2013. Resource Report, Loulo, Mali. Randgold Resources Limited. Internal Company Report. Dated 31st December 2013.

Kleynhans, J., 2013. Resource Report, Gounkoto, Mali. Randgold Resources Limited. Internal Company Report. Dated 31st December 2013.

Loulo Annual Closure Cost Assessment 2017

Loulo Closure Liability 2017

Loulo Gold Mine Tailings assessment – Change order 04, Report11-1900-1049 (CO4), Golder Associates, Sudbury, Ontario, Canada, March 2014.

Loulo Gold Mine Yalea and Gara MRMM, Dempers and Seymour Pty Ltd, Perth, Australia, February 2015.

Loulo Incident Register 2018 Update

Loulo Logging and Mapping Review, Dempers and Seymour Pty Ltd, Perth, Australia, May 2014.

Loulo Map 3D Life of Mine Stress Modelling, Dempers and Seymour Pty Ltd, Perth, Australia, December 2014

Loulo Mining Complex: Hydrogeological review and water management optimisation - Stage 1, Artois Consulting, December 2015.

Loulo Open Pit Ore Reserve Statement 2014, Compiled by Shaun Gillespie, Randgold Resources Ltd, 31 December 2017.

Loulo Optimisation Geotech Closeout_20140702, Dempers and Seymour Pty Ltd, Perth, Australia, July 2013.

Loulo Paste Fill optimisation- December 2014 –V03 (Power point), Gabriel Kone, Loulo, Randgold Resources Ltd, 31st January 2015.

Loulo UG Geotechnical review – Gara and Yalea, Dempers and Seymour Pty Ltd, Perth, Australia, May 2013.

18^th September 2018 Page 312

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Loulo Underground – Drill and Blast Review, Blast Management International Pty Ltd, Australia, June 2014.

Loulo Ventilation Review (Part A) – VUMA Modelling Review, OZvent Pty Ltd, Brisbane, Australia, August 2014.

Loulo Water Balance Jan 2018.

Loulou EMS Audit report Feb 2017.

NQA audit full EMS Report Dec 2017.

On site Tailings assessment, Report11-1900-1049 (4000) R1, Golder Associates, Sudbury, Ontario, Canada, June 2014.

Open Pit Slope Design for the Baboto Pits, MineNet, April 2017.

RAP Audit baseline April 2016.

Report on heat flow modelling of Gara based on increased tonnage, Boet Van Der Vyver, Loulo, Randgold Resources Ltd, September 2014.

Report on heat flow modelling of Yalea based on 135 kt, Boet Van Der Vyver, Loulo, Randgold Resources Ltd, October 2014.

Report on heat flow models of both Gara and Yalea, Boet Van Der Vyver, Loulo, Randgold Resources Ltd, June 2014.

SRK 2009, IndependentNI43-101 Technical Report on the Loulo Gold Mine, Randgold Resources, Mali, Project No 399377, SRK Consulting, Illovo, South Africa, September 2009.

Stope barricade design. Plan, elevation and details, Project No11-1900-1049, Drawing No 1301, Golder Associates, Sudbury, Ontario, Canada, 18 July 2014.

Yalea Crusher & CD Model Report, Dempers and Seymour Pty Ltd, Perth, Australia, January 2015.

Yalea South Upper Geotechnical Review -, Dempers and Seymour Pty Ltd, Perth, Australia, March 2015.

Yalea stoping geometry and backfill strength report, Dempers and Seymour Pty Ltd, Perth, Australia, November 2014.

Yalea stoping geometry and backfill strength report, Dempers and Seymour Pty Ltd, Perth, Australia, November 2014.

18^th September 2018 Page 313

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

28 Date and Signature Page

This report titled “Technical Report on the Loulo-Gounkoto Gold Mine Complex, Mali”, effective 31st December 2017 and dated 18th September 2018, was prepared and signed by the following authors:

		(Signed & Sealed) “Rodney B. Quick”
Dated at London, UK		Rodney B. Quick MSc, Pr. Sci.Nat,
18thSeptember 2018		Group General Manager of Evaluations
		Randgold Resources Ltd.
		(Signed & Sealed) “Simon P. Bottoms”
Dated at London, UK		Simon P. Bottoms, CGeol, MGeol, FGS,
18thSeptember 2018		MAusIMM
		Group Mineral Resource Manager
		Randgold Resources Ltd.
	��	(Signed & Sealed) “Richard Quarmby”
Dated at London, UK		Richard Quarmby, BSc, Pr Eng, C Eng,
		MSAIChE, MIoMMM, MBA
18thSeptember 2018		Group Metallurgist
		Randgold Resources Ltd.
		(Signed & Sealed) “Derek Holm”
Dated at Perth, Australia		Derek Holm, BSc FSAIMM,
18thSeptember 2018		Senior Mining Engineer
		Roscoe Postle Associates
		(Signed & Sealed) “Graham E. Trusler”
Dated at Johannesburg, SA		Graham E. Trusler, MSc, Pr Eng, MIChE,
		MSAIChE
18thSeptember 2018		Chief Executive Officer
		Digby Wells and Associates Pty Ltd.

18^th September 2018 Page 314

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

29 Certificate of Qualified Persons

29.1 Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM

I, Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM as, an author of this report entitled “Technical Report on the Loulo-Gounkoto Gold Mine Complex, Mali”, effective 31st December 2017 and dated 18th September 2018, do hereby certify that:

1. I am the Group Mineral Resource Manager and an Officer of Randgold Resources Limited 3rd Floor, Unity Chambers, 28 Halkett Street, St. Helier, Jersey, Channel Islands.

2. I graduated with a Masters of Geology degree from the University of Southampton, United Kingdom in 2009.

3. I am a Chartered Geologist registered (1023769) with the Geological Society of London. I am also a current Member of AusIMM. I have worked as a geologist continuously for nine years since my graduation from University.

4. I have read the definition of “Qualified Person” set out in National Instrument43-101 (NI43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI43-101.

5. I have been involved at Loulo-Gounkoto since 2012 and regularly visit the site. I most recently visited Loulo-Gounkoto from 17th to 24^thAugust 2018.

6. I am responsible for the preparation of sections 2 to 12, 14, and 23 to 24 of this Technical Report. I share responsibility with myco-authors for sections 1, 25, 26, and 27.

7. I am not independent of the issuer applying the test set out in Section 1.5 of NI43-101 since I am a full time employee at Randgold Resources Limited.

8. I have had prior involvement with the property that is the subject of the Technical Report. I am a full time employee at Randgold Resources Limited and I have been involved with Loulo-Gounkoto since 2012.

9. I have read NI43-101 and Form43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the date of this certificate, to the best of my knowledge, information, and belief, the sections 1 to 4, 6 to 12, 14 and 23 to 27 of this Technical Report contains all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 18th day of September 2018.

(Signed & Sealed) “Simon P. Bottoms”

Simon P. Bottoms, CGeol, MGeol, FGS, MAusIMM

18^th September 2018 Page 315

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

29.2 Rodney B. Quick MSc, Pr. Sci.Nat

I, Rodney B. Quick MSc, Pr. Sci.Nat, as an author of this report entitled “Technical Report on the Loulo-Gounkoto Gold Mine Complex, Mali”, effective 31st December 2017 and dated 18th September 2018, do hereby certify that:

1. I am the Group General Manager of Evaluations and an Officer of Randgold Resources Limited 3rd Floor, Unity Chambers, 28 Halkett Street, St. Helier, Jersey, Channel Islands.

2. I graduated with a Bachelor of Science Honours degree in Geology from the University of Natal Durban, South Africa in 1993, and with a Master of Science degree in Geology from the Leicester University, United Kingdom in 2000.

3. I am a Professional Natural Scientist registered (400014/05) with the South African Council for Natural Scientific Professions (SACNASP). I am a current Member of SACNASP. I have worked as a geologist continuously 24 years since my graduation from University.

4. I have read the definition of “Qualified Person” set out in National Instrument43-101 (NI43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI43-101.

5. I have been involved at Loulo-Gounkoto since 2000 and regularly visit the sit e. I most recently visited Loulo-Gounkoto from 30^thJuly to 3^rdAugust 2018.

6. I am responsible for the preparation of sections 19 and 22 of this Technical Report. I share responsibility with myco-authors for sections 1, 21, 25, 26, and 27.

7. I am not independent of the issuer applying the test set out in Section 1.5 of NI43-101 since I am a full time employee at Randgold Resources Limited.

8. I have had prior involvement with the property that is the subject of the Technical Report. I am a full time employee at Randgold Resources Limited and I have been involved with Loulo-Gounkoto since 2000.

9. I have read NI43-101 and Form43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the date of this certificate, to the best of my knowledge, information, and belief, the sections 1, 2, 3, 19, 21, 22, 24, 25, 26 and 27 of this Technical Report contains all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 18th day of September 2018.

(Signed & Sealed) “Rodney B. Quick”

Rodney B. Quick, MSc, Pr. Sci.Nat

18^th September 2018 Page 316

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

29.3 Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA

I, Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA, as an author of this report entitled “Technical Report on the Loulo-Gounkoto Gold Mine Complex, Mali”, effective 31st December 2017 and dated 18th September 2018, do hereby certify that:

1. I am the Group Metallurgist—Projects and an Officer of Randgold Resources Limited 3rd Floor, Unity Chambers, 28 Halkett Street, St. Helier, Jersey, Channel Islands.

2. I graduated with a BSc chemical engineering degree from the University of the Witwatersrand in 1985 and earned a Master of Business Administration degree in 2005.

3. I have been registered, no. 910237 as a Professional Engineer (Pr Eng) with the Engineering Council of South Africa since 1991 and in 2010 was accepted to the UK equivalent institution i.e. Chartered Engineer with the Engineering Council UK (C Eng), no. 580441. Further, I have been a Member, no. 1361, of the South African Institution of Chemical Engineers (SAIChE) since1989, and am also a registered Member, no. 454225, of the Institute of Materials, Minerals and Mining (IoMMM) UK. I have worked as an engineer continuously since my graduation from University in 1985.

4. I have read the definition of “Qualified Person” set out in National Instrument43-101 (NI43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI43-101.

5. I have been involved at Loulo-Gounkoto since 2016 and regularly visit the site. I most recently visited Loulo-Gounkoto from 13^thto 16^thJune 2018.

6. I am responsible for the preparation of sections 13, 17 and 18 of this Technical Report. I share responsibility with myco-authors for sections 1, 21, 25, 26, and 27.

7. I am not independent of the issuer applying the test set out in Section 1.5 of NI43-101 since I am a full time employee at Randgold Resources Limited.

8. I have had prior involvement with the property that is the subject of the Technical Report. I am a full time employee at Randgold Resources Limited and I have been involved with Loulo-Gounkoto since 2016.

9. I have read NI43-101 and Form43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the date of this certificate, to the best of my knowledge, information, and belief, the sections 13, 17 and 18 of this Technical Report contains all scientific and technical information that is required to be disclosed to make the Technical Report not misleading.

Dated this 18th day of September 2018.

(Signed & Sealed) “Richard Quarmby”

Richard Quarmby, BSc, Pr Eng, C Eng, MSAIChE, MIoMMM, MBA

18^th September 2018 Page 317

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

29.4 Derek Holm, BSc, FSAIMM

I Derek Holm, BSc, FSAIMM, as an author of this report entitled “Technical Report on the Loulo-Gounkoto Gold Mine Complex, Mali”, effective 31st December 2017 and dated 18th September 2018, do hereby certify that:

1. I am a Senior Mining Engineer with Roscoe Postle Associates UK (RPA UK), of London, UK.

2. I graduated from University of Witwatersrand in 2000 with a BSc in Mining Engineering.

3. I am a Member of the South African Institute of Mining and Metallurgy (SAIMM) (no. 702981). I have worked as an engineer continuously since my graduation from University in 2000.

4. I have read the definition of “Qualified Person” set out in National Instrument43-101 (NI43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI43-101. I have over 18 years of experience within the mining industry in technical, management, and consulting positions.

5. I visited Loulo-Gounkoto 12^thto 14^thSeptember 2018.

6. I am responsible for the preparation of sections 15 and 16 of this Technical Report. I share responsibility with myco-authors for sections 1, 25, 26, and 27

7. I am independent of Randgold Resources Limited and related companies applying the test set out in Section 1.5 of NI43-101.

8. I have had no prior involvement with the property that is the subject of the Technical Report.

9. I have read NI43-101 and Form43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the effective date of this report, to the best of my knowledge, information, and belief, sections 15, 16 and 18, and parts of sections 1, 3, 25, 26, and 27 of this Technical Report for which I am responsible contains all scientific and technical information that is required to be disclosed to make this Technical Report not misleading.

Dated this 18th day of September 2018.

(Signed & Sealed) “Derek Holm”

Derek Holm, BSc, FSAIMM

18^th September 2018 Page 318

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

29.5 Graham E. Trusler, MSc Pr Eng, MIChE, MSAIChE

I, Graham E. Trusler, Pr Eng, MIChE, MSAIChE, as an author of this report entitled “Technical Report on the Loulo-Gounkoto Gold Mine Complex, Mali”, effective 31st December 2017 and dated 18th September 2018, do hereby certify that:

1. I am the CEO of Digby Wells and Associates Pty Ltd., of Johannesburg, South Africa.

2. I graduated with a Master of Chemical Engineering degree from the University of KwaZulu-Natal, South Africa, in 1988.

3. I have been registered as a Professional Engineer (Pr Eng) (no. 920088) with the Engineering Council of South Africa since 1992. Further, I have been a Member of the South African Institution of Chemical Engineers (SAIChE) since1994. I am also registered as a Chartered Chemical Engineer with the Institution of Chemical Engineers, is a member of the Water Institute of South Africa and a lifetime member of the American Society of Mining and Reclamation. I have worked as an engineer continuously from 1990.

4. I have read the definition of “Qualified Person” set out in National Instrument43-101 (NI43-101) and certify that by reason of my education, affiliation with a professional association (as defined in NI43-101) and past relevant work experience, I fulfil the requirements to be a “Qualified Person” for the purposes of NI43-101. I have over 30 years of experience within the mining industry in metallurgical production, research, and environmental issues.

5. I have been involved at Loulo-Gounkoto since 1998 and regularly visit the site. I most recently visited Loulo-Gounkoto from 18^thto 20^thApril 2018.

6. I am responsible for the preparation of section 20 of this Technical Report. I share responsibility with myco-authors for sections 1, 25, 26, and 27.

7. I am independent of Randgold Resources Limited and related companies applying the test set out in Section 1.5 of NI43-101.

8. I have had prior involvement with the property that is the subject of the Technical Report.

9. I have read NI43-101 and Form43-101F1 and the Technical Report has been prepared in compliance with that instrument and form.

10. As of the effective date of this report, to the best of my knowledge, information, and belief, sections 20, and parts of sections 1, 2, 3, 25, 26, and 27 of this Technical Report for which I am responsible contains all scientific and technical information that is required to be disclosed to make this Technical Report not misleading.

Dated this 18th day of September 2018.

(Signed & Sealed) “Graham E. Trusler”

Graham E. Trusler, MSc, Pr Eng, MIChE, MSAIChE

18^th September 2018 Page 319

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

30 Appendix

30.1 Appendix 1 – JORC 2012 Edition – Table 1

The following table provides a summary of important assessment and reporting criteria used at the Loulo-Gounkoto Gold Complex for the reporting of Mineral Resources and Ore Reserves in accordance with the Table 1 checklist in The Australasian Code for the Reporting of Exploration Results, Mineral Resources and Ore Reserves (The JORC Code, 2012 Edition). Criteria in each section apply to all preceding and succeeding sections.

Section 1. Sampling Techniques and Data

JORC (2012) Code Checklist of Assessment and Reporting Criteria

Criteria	Commentary
Sampling techniques	●   Diamond drill (DD) core samples are taken within geological units and are normally between 0.8 m and 1.20 m long. Diamond drilling has been primarily undertaken by Boart Longyear Canada. The core is photographed before being halved with a diamond saw. Half of the core is submitted for sampling and the other half is stored for future reference.
	●   Reverse circulation (RC) samples are sampled on one metre intervals. The RC samples are ether cone split or riffle split on the rig. RC drilling is completed by Boart Longyear Canada, except for some limited Gounkoto grade control drilling that is completed by DCS Mali. Rotary air blast (RAB) drilling has also been undertaken previously, but is only used for geological modelling and not for resource estimation
	●   Outcrop, trench, and soil samples are also used for early stage exploration. Trench data with exposure cuts into saprolite is used for estimation and the sampling technique is the same as RC drilling.
	●   A February 2015 Mineral Resource audit by QG Consulting (QG) deemed data collection at the Project follows industry standard practices
Drilling techniques	●   Diamond drilling is utilised for exploration and resource development. PQ rods (85.0 mm) and HQ rods (63.3 mm) are used in the weathered saprolite (oxides) with NQ (47.6 mm) rods are used in the unweathered rock. Diamond core is stored in core trays that are marked up and logged before transfer to the core storage.
	●   Reverse circulation is no longer used at Yalea and Gara underground operations’ but is still used at Gounkoto and other open pit satellites.
	●   RAB drilling has been used for first pass exploration and sterilisation, but it not used in resource estimation
	●   All boreholes have been down-hole surveyed using Reflex EZ-Trac instrument every 25 m depth which measures the dip and azimuth. A reflex gyro instrument is used in any zones with magnetic interference.

18^th September 2018 Page 320

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
Drill sample recovery	●   Core recovery is measured in the field and during detailed logging.
	●   RC sample recovery is measured by weighing the total dried weight of sample collected over each one metre interval and comparing this to the theoretical expected weight for the lithological unit and weathering type.
	●   Historical studies at Gounkoto have confirmed there to be no significant grade bias through sample recovery rates.
Logging	●   Diamond drill core is logged geologically including weathering, mineralisation, alteration, structure, lithology, and redox. The paper logs are transcribed into the central database using a digital data entry template after verification.
	●   Geotechnical logging is only performed on holes specifically drilled for geotechnical assessment.
	●   RC chip samples are logged in the same fashion as the drill core, however the samples are logged on one metre intervals due to the sampling method.
Sub-sampling techniques and sample preparation	●   Half core is used whenever possible. Quarter core is only used when there is a requirement for further assays or other analysis types need to be performed on the same interval. Drill core length for each assay is determined by the geology and alteration.
	●   RC samples are collected and passed through two riffle splitters to create 2 kg to 3 kg samples. Some minor grade control drilling in Gounkoto uses an on rig cone splitter. Any wet samples are dried before being split.
Quality of assay data and laboratory tests - Loulo	●   All samples are analysed by SGS Loulo Laboratory, and onsite laboratory managed and self-certified by SGS laboratory personnel. All samples are analysed using fire assay (FA550) with submitted sample weights above 2 kg and 150 g for Certified Reference Materials (CRM).
	●   CRM standards, blanks, field duplicates and umpire assay are used for QA/QC analysis. The QA/QC sample ratio during the reporting period showed CRM sample insertion at 1 in 18 (5.6%), blank sample insertion at 1 in 18 (5.6%) and field duplicate sample insertion at 1 in 36 (2.8%).
	●   During 2017, 370 sample pulps duplicates from Loulo along with CRMs were sent from SGS Loulo to AMTEL laboratory in Canada, for umpire analysis. The umpire results indicate that there is no material bias in the primary laboratory (SGS Loulo).
	●   Coarse blank samples are composed of a barren sandstone material sourced 20 km from Loulo, which is considered to be a representative matrix of the Loulo host rock lithological units.
	●   During 2017, 730 blank samples were submitted where 100% assayed were within five times the detection limit. Overall the performance of the blanks was deemed as good.
	●   Randgold use 12 different CRM standards. During 2017 a total of 731 CRM standards were submitted to SGS Loulo. The CRMS indicated that there was no significant bias at SGS Loulo.
	●   Overall the results returned are acceptable and show no major areas of concern.
	●   The 2015 external audit by QG concluded that the sample preparation and assay facility at SGS Loulo follows industry standard methods with a good emphasis on sample tracking and QA/QC.

18^th September 2018 Page 321

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
Quality of assay data and laboratory tests - Gounkoto	●   All samples are analysed for the Gounkoto Gold Mine by SGS Loulo Laboratory which is managed and self-certified by SGS laboratory personnel. All samples are analysed using fire assay (FA550) with submitted sample weights above 2 kg and 150 g for CRMs.
	●   At Gounkoto Gold Mine, CRM standards, blanks, field duplicates and umpire assay are used for QA/QC analysis. The QA/QC sample ratio during the reporting period showed CRM sample insertion at 1 in 18 (5.6%), blank sample insertion at 1 in 18 (5.6%) and field duplicate sample insertion at 1 in 36 (2.8%).
	●   During 2017, 261 sample pulps duplicates from Gounkoto along with CRMs were sent from SGS Loulo to AMTEL laboratory in Canada, for umpire analysis. The umpire results indicate that there is no material bias in the primary laboratory (SGS Loulo).
	●   Coarse Blank samples are composed of a barren sandstone material, which is considered to be a representative matrix to the Gounkoto host rock lithological units.
	●   During 2017 2,170 blank samples were submitted where 99.95% assayed within 10% of acceptable limits. Overall the performance of the blanks was deemed as good.
	●   During 2017 a total of 2172 CRM standards of 10 different types were submitted to SGS Loulo. Overall of CRM improved significantly in 2017 due to the introduction of CRMs sourced from OREAS Australia, as CRMs did show very good homogeneity.
	●   Overall the results returned are acceptable and show no major areas of concern.
	●   The 2015 external audit by QG concluded that the sample preparation and assay facility at SGS Loulo follows industry standard methods with a good emphasis on sample tracking and QA/QC.
Verification of sampling and assaying	●   Twin drilling at Loulo (Yalea and Gara) was utilised as part of the 2002 feasibility studies prior to construction of UG mine development. These comparisons have shown that although there can be variations in grade, as expected, the broad intercepts and relative grade of the intersections are comparable across the twins.
	●   Twin drilling at Gounkoto have been utilised to compare seven separate drill holes across the MZ and HW. These comparisons have shown that there can be significant variations in grade, as expected, from the local grade variability in Gounkoto MZ1, MZ2 and MZ3. However, the broad intercepts and relative grade of the intersections are generally comparable across the twins except in the FW finger zone where there are dramatic variations in geometry within a very close spacing (less than 5 m).
	●   Twin drilling at all other mineral resources is included as part of the resource definition drilling.
	●   All forms of Project data are stored secured in industry standard Maxwell Datashed SQL database for optimal validation through constraints, library tables, triggers, and stored procedures. Any data that fails validation is rejected and stored in a buffer table awaiting correction.
	●   A custom MS Access front end application has been designed for data entry, reporting, and viewing which connects via an ODBC connection to the SQL database. Data which is inserted utilises the data validation procedures from the SQL database.
	●   Any site databases link back to the main database for information retrieval via ODBC
	●   Assay data is imported directly from laboratory assay certificates and validated. Assay data are stored in a normalised format and multiple assays are stored for each sample. Ranking of different assay formats is performed automatically so that one assay result is displayed in the master assay table. Any change to the rankings held within a lookup table must be approved by the onsite Database Manager.

18^th September 2018 Page 322

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
Location of data points	●   Grade control drill hole collar positions are marked out by surveyors and all drill holes collar positions are surveyed using DGPS on surface.
	●   Underground drill collars, as well as back sights and foresights, are surveyed using total station underground survey instruments, and marked on the drift walls, by the Loulo Mine Surveyors.
	●   All drill holes are down-hole surveyed using a multi shot Reflex EZ-Trac or a conventional Gyro surveying tool which takes measurements of azimuth and dip every 25 m. Some drill holes for the 2002 underground feasibility study were measured using a Gyro every 5 m.
	●   Loulo-Gounkoto uses UTM coordinates WGS84 Zone 29N.
Data spacing and distribution - Loulo	●   The data spacing for resource classification is constrained for each deposit and is combined with other classification specifications such as minimum samples used to estimate and style of mineralisation.
	●   For Yalea in general, Measured material requires grade control (GC) infill drilling at a range of spacing of 25 m to 30 m or less, Indicated material must have a minimum drill spacing of 70 m and Inferred material has a drilling density generally greater than 70 m to 120 m.
	●   For Gara in general, Measured Mineral Resources require GC infill drilling at a range of spacing of less than 30 m, Indicated resources must have a minimum of between 30 m to 80 m spacing and Inferred resources has a drilling density of 80 m to 120 m.
	●   For Baboto in general, Measured material requires GC infill drilling at a range of spacing of 5 m by 10 m, Indicated and Inferred material must have a minimum drill hole spacing of 80 m spacing although with other classification criteria changing depending on the classification.
	●   Sample compositing is applied, prior to top cutting, for samples that are used in the resource estimation. This is currently on a 2 m composite as this is a multiple of the mode sample length for Yalea and Gara. The Satellite deposits including Baboto use a 1 m composite which is the mode sample length.
Data spacing and distribution - Gounkoto	●   The data spacing for resource classification is clearly constrained for the deposit. Amongst other classification specifications, in general, Measured material requires a minimum of GC infill drilling at a spacing of 12.5 m x 12.5 m with tighter spacing of 6m both along strike and across dip in the FW finger zones.
	●   Indicated material within the Gounkoto super pit must have a minimum of drilling with at least 30 m spacing and the UG Indicated material has a spacing of approximately 40 m by 30 m due to restrictions in drill locations.
	●   Inferred material has a drilling spacing between 50 m and 100 m.
	●   Sample compositing is applied, prior to top cutting, for samples that are used in the resource estimation. This is currently on a 1.0 or 2.0 m composite as this is a multiple of the modal sample length.
Orientation of data in relation to geological	●   At Yalea, the majority of surface drilling is from the East dipping at a bearing of 60° toward the West.
	●   At Gara the orientation of drilling is much more variable in response to the folded geometry of the mineralised units. Exploration holes are drilled from the east and west, dipping at around 60oto ensure that drilling is perpendicular to mineralisation.
	●   Underground orientations depend on available drilling positions.

18^th September 2018 Page 323

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
structure - Loulo	●   All surface holes are generally drilled perpendicular to the strike of the mineralisation on the hanging wall / down dip side of the deposit.
Orientation of data inrelation to geological structure - Gounkoto	●   Most exploration holes are drilled from the east, dipping to the west at around 60oto ensure that drilling is perpendicular to mineralisation e.g. MZ1.
	●   Grade control drilling has different orientations depending on the ore zone that is being intercepted. Footwall finger zones of MZ1 are drilled from the west to ensure that the high-grade ‘finger’ zones are intersected across their dip. MZ2 and MZ3 are drilled from the east
Sample security	●   Samples on the rigs are bagged and tied with custom Loulo or Gounkoto tags as well as being weighed and documented. The samples are stored in a secure facility. Samples are analysed at SGS Loulo laboratory which is on the Loulo Permit. When samples are required to be prepared/analysed at other laboratoriesoff-site, SGS provide secure transportation for the samples.
Audits or reviews	●   An external audit on the mineral resource and input data procedures was completed in February 2015 by QG. This report stated that sampling procedures for RC and diamond drilling were considered to be good practice.
	●   Some minor improvements for the collection of bulk density measurements (use of recording balance and certified weights) were suggested and were implemented during 2016.
	●   Outcomes of the external audits have been acted upon where relevant.

18^th September 2018 Page 324

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Section 2. Reporting of Exploration Results

(Criteria listed in the preceding section also apply to this section.)

Criteria
Mineral tenement and land tenure status - Loulo	●   The Loulo deposits are located 22.5 km north of Randgold Resource’s Gounkoto Gold Mine, within the Loulo Exploitation Permit (Loulo Permit). This Permit is in Western Mali approximately 350 km west of Bamako and 220 km south of Kayes. The deposit lies within the Loulo Permit which is held by Société des Mines de Loulo SA of which there is an 80/20 split between Randgold Resources and the State of Mali.
	●   The Loulo Establishment Convention sets out under the 1991 Mining Code that a six percent royalty payable on revenues along with a corporate tax rate on profits of 30% with a minimum of 0.75% on gross revenues is due to the State.
Mineral tenement and land tenure status - Gounkoto	●   The Gounkoto deposit is located 22.5 km south of Randgold Resource’s Loulo Gold Mine. This Permit is in Western Mali approximately 350 km west of Bamako and 220 km south of Kayes. The deposit lies within the Gounkoto Exploitation Permit (Gounkoto Permit) which is held by Société des Mines de Gounkoto SA which is held 80% by Randgold and 20% by the sate of Mali. The Gounkoto Permit incorporates Faraba and is located to the south of the current Loulo Permit It is valid for a period of 30 years from August 2012.
	●   The Loulo Establishment Convention sets out under the 1991 Mining Code that a six percent royalty payable on revenues along with a corporate tax rate on profits of 30% with a minimum of 0.75% on gross revenues is due to the State.
Exploration done by other parties	●   The Gara deposit (previously called Loulo 0 deposit) was discovered in 1981 by the Syndicat Or joint venture.
	●   In 1992, BHP Minerals Mali (BHP) entered into an agreement with SOMILO (Société des Mines de Loulo), the joint venture wholly owned by the Republic of Mali and SOREM (a 100% subsidiary of BRGM). BHP undertook exploration mostly focused on Gara.
	●   BHP Minerals Mali delineated a Mineral Resource on the Gara deposit prior to 1996. The Mineral resource contained 3.48 Mt at 4.43 g/t Au of Open Pit Indicated material, 0.87 Mt at 9.7 g/t of underground Indicated material and 1.0 Mt at 10.0 g/t Inferred material. An additional 1.42 Mt at 4.63 g/t of Inferred material was identified at outlying satellite deposits
	●   Randgold acquired BHP Minerals Mali in 1996 and changed its name to Randgold Resources (Mali) Ltd. Following the discovery of the Yalea deposit a series of feasibility studies were undertaken.
	●   Gounkoto is a greenfields discovery by Randgold and thus has not had any other external parties working on it.
Geology - Loulo	●   Loulo is located within the Kedougou-Kenieba erosional inlier. The inlier is unconformably overlain by Upper Proterozoic sandstones towards the east and further south. The Kenieba inlier contains several significant gold deposits including Sadiola, Yalea, Segala, Tabakoto, and Gounkoto deposits in Mali, and Sabodala in Senegal. The Senegal-Mali shear marks a major break in the geology from shelf carbonates in the west to the sedimentary sequences of the Kofi formation in the east. This geological setting is the primary host of mines in Burkina Faso, Ghana, Mali, Niger, and Senegal.

18^th September 2018 Page 325

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria
	●   At Yalea, the main mineralised body is a hosted by the Yalea Shear, where it is intercepted by the Yalea Structure. The Yalea Shear is a brittle-ductile, north-south striking, mineralised fault that transects the Yalea Structure, which is a complex, north to NNE striking shear zones. The Yalea mineralisation is predominantly hosted in hydrothermally brecciated argillaceous pink quartzites situated. A higher grade ‘Purple Patch’ zone is observed in a dilatational strain transfer zone formed as the western dip of the upper mineralised system steepens, forming hydraulic breccias. Economic levels of gold mineralisation are almost exclusively associated with paragenetically late sulphide veins, breccias and zones of massive sulphides. There is a strong correlation between sulphide intensity and gold mineralisation with the dominant sulphide phases consisting of pyrite (abundant), arsenopyrite and minor chalcopyrite. Yalea mineralisation, remains open at depth and to the south with potential for significant high-grade extensions.
	●   Gara (previously known as Loulo 0) is hosted within an intensely tourmaline greywacke unit which outcrops on surface due to its high resistance to weathering. The geometry of the mineralisation is subjugated by the strike slip shearing on Senegal-Mali shear. This shearing has resulted in folding, fracturing, brecciation, and subsequent development of a quartz-carbonate vein stockworks within the brittle-ductile tourmaline altered greywacke forming what has been termed quartz tourmaline (QT) unit. On the deposit scale the upper limb of this fold dips has a westerly dip, whereas the lower limb dips east. The distribution of gold grade in long section reflects the varying degrees of fracturing, brecciation, and subsequent development of a quartz-carbonate vein stockworks from multiple generations of folding. Gold mineralisation is strata-bound and hosted predominantly within the quartz-tourmaline stockwork veins, which are enveloped within footwall greywackes and hanging wall sandstone. In the open pit area, the high-grade mineralisation is concentrated along the sub- horizontal fold hinge axes, whereas within the underground area, high-grade mineralisation plunges shallowly southward, parallel to the large scale open warp fold axis. The sulphide assemblages predominantly consist of disseminated auriferous pyrite with minor chalcopyrite, scheelite, and nickeliferous sulphides
	●   Baboto is a shear hosted deposit situated along a north-south striking shear structure located approximately 14 km NNE from the Yalea deposit. Baboto is dominated by a thick sequence of metasediments and structural breccias. The main shear zones are vertical to steeply west dipping at Baboto South and sub vertical in Baboto Centre. Gold mineralisation is mainly associated with the finely disseminated pyrite occurring in the brittle-ductile shear breccias, which generally have a lensoidal shape defined by a series ofsub-parallelN-S shears that follow key lithological contacts.
	●   Loulo 3 is located 4 km NNE of the Yalea mine. Loulo 3 consists of three mineralised zones: a NNW trending main zone (MZ1) which is situated on the Loulo 3 structure and is transected by the NNE striking main zone (MZ2), which is situated on the Yalea structure, and the third small sub parallel NW striking footwall zone. Mineralisation consists of a mixture of quartz and hematite veinlets hosted in a zone of silica-carbonate alteration within local tourmaline alteration in the south. The distribution of high-grade zones is controlled by the narrowing of the host stratigraphy package, which focusses strain and fluid flow, causing the hematite rich Yalea Structure to interact with the silica- carbonate Loulo 3 Structure particularly within MZ2. Gold bearing sulphides predominantly consist of pyrite and arsenopyrite.
	●   Gara West is located 200 m west of the pit at Gara and is characterised by predominantly shear and breccia hosted mineralisation within a medium to coarse grained sandstone unit that is variably altered with tourmaline, chlorite, and silica-carbonate. The sandstone hosts four mineralised lodes striking NNE and dipping moderately westward. The gold mineralisation is strata bound as it has been preferentially

18^th September 2018 Page 326

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria
	altered with tourmaline (and silica-albite), due to the increased porosity of the protolith, relative to the bounding limestone in both the hanging wall and footwall.
	●   Other minor satellite deposits are present within the Loulo Permit, these exhibit similar geological characteristics to the other major deposits outlined above.
Geology – Gounkoto	●   Like Loulo, the Permit is located within the Kedougou-Kenieba erosional inlier in a sheared Birimian Greenstone belt.
	●   Gounkoto is a large NNW trending shear zone, with a complex assemblage of ductile shear breccias, shears, and faults characterised by a stepped geometry, with wider zones of mineralisation generally seen on the NW trending structures and narrower zones on the north- south trending structures. This is believed to be related to dilation across these structures in a strong sinistral strain environment. The mineralisation is generally hosted in a siliceous ‘Rose Quartzite’ (QR) unit. The mineralisation is subdivided based on the structural and lithological characteristics.
	●   The Faraba deposit strikes NNW and is comprised of several zones of gold mineralisation hosted within and along the contacts of north- south striking, coarse grained, gritty sandstone units (lithic wackes) in a package of sheared argillaceous sediments. Lithologic layering (transposed bedding) dips steeply westward; however, the mineralised zones dip steeply to the east. The mineralisation terminates where the Faraba Structure meets the argillite units on either side of the sandstones. The resulting mineralisation occurs as numerous silica- carbonate and secondary iron oxide altered sub vertical panels with narrow east-west dimension, each containingsub-horizontal to shallow plunging zones of higher grade. Gold mineralisation is dominantly hosted by pyrite, with local magnetite, chalcopyrite, arsenopyrite and pyrrhotite.
Drill hole Information	●   Exploration at Loulo-Gounkoto is focussed on advancing both brownfields and greenfields targets. Brownfields exploration involves testing underground and open pit targets for high-grade mineralisation based on the structural model.
	●   No exploration results have been announced in this report.
Data aggregation methods	●   Individual exploration results are seldom reported for Loulo-Gounkoto. Where applicable any results are capped to the same grades as those used for the relevant geological domain defined within the existing Mineral Resource.
Relationship between mineralisation widths and intercept lengths	●   No exploration results are reported in this document, but drilling is orientated as perpendicular to orebody as possible. All drill results are reported with the true thickness along with down hole depths.

18^th September 2018 Page 327

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria
Diagrams	●   No exploration results are reported in this document.
Balanced reporting	●   No exploration results are reported in this document.
Othersubstantive exploration data	●   Recent geophysical surveys have been integrated with updated geological maps to develop a tectonostratigraphy of the Kofi Series rocks at Loulo-Gounkoto, to improve the understanding of the controls to gold mineralisation and the regional geologic architecture. This Project- wide geologic framework is driving are-assessment of exploration work to date as part of greenfields target generation.
	●   The current exploration concept is proved to be effective with both the discovery of Gounkoto and the successful replenishment of depleted Mineral Resources and Ore Reserves.
Further work - Loulo	●   The key aims of the 2018 Loulo exploration programme are to replace 2018 depletion and discover a stand-alone resource or significant satellite to increase Mineral Resources and LOM of Loulo-Gounkoto.
	●   Exploration is targeting the northern side of the deeper Yalea intersections panels, if there is a positive economic scope. Additionally, exploration will target the central conversion target and conceptual targets along strike of the Yalea Transfer Zone.
	●   Advanced grade control drilling of Yalea that commenced in 2017 is planned to continue into 2018 to target the Yalea Far South Extension.
	●   Loulo 3 drilling is designed to target plunge extensions of MZ2 shoot at near surface and at depth and to test the variability in the mineralisation thickness and grades across at a range of drill spacings.
Further work - Gounkoto	●   As at Loulo, the exploration team is focused on the primary objective of replacing 2018 depletion and discovering additional resources to increase the inventory of Loulo-Gounkoto.
	●   During 2018, exploration will target a southerly plunging shoot of chlorite and subordinate haematite and tourmaline alteration which is known to contain higher grade mineralisation.
	●   Follow-up RC drilling is also planned to target a HW Hematite Structure, in the eastern portion of the Faraba open pit.

18^th September 2018 Page 328

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Section 3. Estimation and Reporting of Mineral Resources

(Criteria listed in section 1, and where relevant in section 2, also apply to this section.)

Criteria	Commentary
Database integrity	●   Data is stored in a Maxwell Datashed SQL database. Data must pass validation through constraints, library tables, triggers, and stored procedures prior to importing. Failed data is either rejected or stored in buffer tables awaiting correction. Assay data is imported directly from laboratory certificates and only fully trained and authorised network users can upload laboratory data. All other MS Access databases on site use ODBC link to retrieve information from the Datashed SQL database thus ensuring that they utilise all SQL database validation mechanisms. A full-time database administrator is employed at site to manage the database, who is overseen by a regional database manager,
	●   Digital logging solutions for use by geologists to prevent transcription errors are planned for implementation in 2019.
Site visits	●   Qualified Person Simon Bottoms CGeol regularly visits the site. During 2017 six separate visits were made to the Loulo and Gounkoto Permits to review exploration programmes and results; Mineral Resource and grade control model updates; in addition to board reviews.
Geological Interpretation	●   Vertical sections are created using Vulcan or Gemcom geological modelling software at regular sections dependent on the drilling density. Sections are orientated depending on the strike of the mineralised deposit. Onsite geologists prepare their interpretations on printouts of the sections, using the drilling, lithological, alteration, mineralisation, structure measurement logging together with assay data from the surface and pit mapping, trenching, RC and diamond drill hole data. From these interpretations, strings representing the outlines of the orebody were digitised. The strings are connected to form a three-dimensional triangulation/wireframe using Vulcan or Gemcom geological 3D modelling software.
	●   The Loulo deposits have beensub-divided into different geological domains. Yalea has been split into 7, Gara into 18, Baboto into 5, Gara West into 4 and Loulo 3 into 3 different geological domains.
	●   Gounkoto has beensub-divided into nine geological domains. There are four main zones (MZ1, 2, 3, 4), two footwall zones (FWFE, FWNE), two hanging wall zones (HW) and P64.
Dimensions - Loulo	●   Yalea has a strike length of 2.7 km, a maximum defined vertical depth of 1,050 m and an average thickness of approximately 25 m The bottom of the open pit resource is around 150 m to 180 m below surface sea level and the underground resources are defined at present to around 830 m surface.
	●   Gara has a strike length of 2.8 km and has a maximum defined vertical depth of 920 m with a thickness between 5 m and 25 m The bottom of the defined open pit resource is around 180 m below surface and the underground resources are defined at present to around 1,010 m below surface.

18^th September 2018 Page 329

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   Baboto South is 1.5 km along strike, a 320 m defined depth, and an average thickness of 10m. Baboto Centre is 700 m along strike, 140 m in depth, and has an average thickness of 15m.
	●   Loulo 3 has a strike length of 1.8 km, ranges in thickness between 6 m and 12 m and is drilled to a vertical depth of 370 m.
	●   Gara West consists of two mineralised zones (Western and Eastern). The western zone has a strike length of 480 m, and a thickness of 9 m. Similarly, the eastern zone is 440 m in strike, 6 m in thickness. Both zones have demonstrateddown-dip continuity of 130 m.
Dimensions - Gounkoto	●   Gounkoto has an along strike length of 2.2 km, a vertical depth of 700 m, and a thickness between 2 m and 40 m. The bottom of theopen-pit resource is-275 m RL in the north and-190 m RL in the south. The underground resources are limited to 670 m depth in the south, 700 m in the centre, and 450 m depth at the northern end.
	●   Faraba has an along strike length of 900m and a thickness of 5 m to 30m. It has a maximum vertical depth of 320 m but mineralisation is truncated in a curvilinear manner where the Faraba Structure meets the argillite units on either side of the sandstones.
Estimation and modellingtechniquesLoulo	●   Each deposit issub-domained separate zones based upon geology, mineralisation, alteration, and grade populations.
	●   Drill data is composited between the domain boundaries on two metre intervals at Yalea and Gala, and one metre intervals at Baboto, Loulo 3 and Gara West. A minimum of half the composite length is permitted at domain boundaries with residuals being discarded
	●   Comparison studies found very little difference between the total sample intervals and contained metals when comparing the raw and composited samples. Residual analysis is undertaken to ensure the discarded data does not affect the estimation.
	●   Multifactor data analysis is completed on each separate geological domain with the aim of identifying any grade outliers and choose an optimum top cap value for each of the orientation domains. The multifactor analysis tools applied include: Histogram plot, Log probability plot, Disintegration analysis, Mean and CV curve to look for the stability point and Metal at Risk (MAR).
	●   A 2015 external audit found these decisions to be supportable and appropriate.
	●   Exploratory Data Analysis (EDA) and variography is undertaken on the estimation data using Snowden Supervisor for determining the search strategies. Where insufficient data is available for variography on a domain, the results of a similar/parent domain are applied and rotations updated to match the domain.
	●   Quantitative Kriging Neighbourhood Analysis (QKNA) is undertaken to aid determining optimum block sizes, sample numbers, ellipsoid volumes, and discretisation
	●   Wireframes and block models are furthersub-domained into exploration domains, advanced grade control and grade control domains which allows for the use of different minimum block sizes and search ellipses. The block sizes used in the separate domains are intended to reflect the average drill hole spacing within the different scenarios.

18^th September 2018 Page 330

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria		Commentary
	●   Block model parameters are listed below. Block models are rotated around the Z axis to match the strike of the mineralisation.
		Block Detail (X/Y/Z)	Yalea	Gara	Loulo 3	Baboto	Gara West
		Parent Block Size (m)	10 by 30 by 20	7.5 by 30 by 15	10 by 20 by 5	5 by 15 by 10	10 by 20 by 5
		GC Domain Block Size (m)	5 by 15 by 10	5 by 20 by 10	5 by 10 by 5	3 by 6 by 2.5	-
		Adv GC Domain Block Size (m)	8 by 24 by 16		-	-	-
		Rotation (°)	90	90	100	90	90
	●   Volume comparisons between the block model and the ore wireframes have shown overall that an appropriate block size and sub-blocking procedure is in place.
	●   Estimation was completed using ordinary kriging (OK) and parent cell estimation; any blocks that were estimated using a negative kriging weight were reset to zero grade.
	●   For Baboto a restriction on high yield outliers was used to limit samples above 5 g/t Au in the exploration area of the Baboto South and 30 g/t Au in the Baboto South GC area in the third pass on the estimation.
	●   In Yalea, Geological domain 9006 a high yield limit is used with samples above 38.41 g/t Au not being used if outside of 45 m by 21 m by 6 m search ellipse.
	●   Block models are depleted against a 5 m Lidar DTM that was generated in 2010, along with annual Lidar scans of open pits and underground development stope scans. Grades are retained in the block model for reconciling to life of mine information
	●   Validation procedures are undertaken on the estimations. These include comparison of mean grades, visual comparisons to composite grades, slope of regression calculations, comparison to nearest neighbour analysis and Swath plots. As expected, Swath plots indicate increased smoothing outside of GC areas where exploration data is more limited. No conditional bias is observed.
Estimation and modelling techniques Gounkoto	●   Gounkoto and Faraba were domained into separate zones along the strike of the deposit. ●   Drill hole intervals that were within the domains were composited to 2.0 m at Gounkoto and 1.0 m at Faraba ●   Multifactor data analysis is completed on each separate geological domain with the aim of identifying any grade outliers and choose an optimum top cap value for each of the orientation domains. Composited samples at Gounkoto were either capped between 1.73 g/t or subjected to high-grade constrains during estimation to restrict outliers. Faraba samples were capped to between 8.65 g/t Au and 20.79 g/t Au.
	●   Exploratory Data Analysis (EDA) and variography is undertaken on the estimation data using Snowden Supervisor for determining the search strategies. Where insufficient data is available for variography on a domain, the results of a similar/parent domain are applied and rotations updated to match the domain.
	●   Quantitative Kriging Neighbourhood Analysis (QKNA) is undertaken to aid determining optimum block sizes, sample numbers, ellipsoid volumes, and discretisation
	●   At Gounkoto the sub domaining of the main ore wireframes into exploration domains and grade control domains allows for the use of different minimum block sizes and search ellipses. The block sizes used in the two separate domains are intended to reflect the average drill hole spacing within both scenarios.

18^th September 2018 Page 331

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   At Gounkoto a parent block size of 15 m by 30 m by 9.9 m (XYZ) was applied to exploration areas, advanced grade control areas used a 7.5 m by 15 m by 4.95 m (XYZ) block size and grade control zones used a 5 m by 10 m by 3.3 m block size. Soft boundaries are applied between the different data density areas that are situated within the same geological domain. ●   At Faraba a single 20 m by 20 m by 10 m (XYZ) block size was sed. ●   At Gounkoto, MZ1 structural domains apply a firm boundary which allows the west dipping and central west domains to use 40% of the search distance to use data from the east dipping domain, but hard bounds the east dipping domain from the other domains to prevent the higher grades in the west and central west dipping domains from being smeared into the lower grade east dipping population. ●   Volume comparisons between the block model and the ore wireframes has shown that an appropriate block size andsub-blocking procedure is being used. ●   Estimations were completed using ordinary kriging and parent cell estimation; any blocks that were estimated using a negative kriging weight were changed to the nearest anisotropic block grade. Domain MZ3HW, was estimated using inverse distance cubed as there was insufficient data to determine kriging parameters and no geologically comparative domain was identified ●   Block models are depleted against a 5 m Lidar DTM that was generated in 2010, along with annual Lidar scans of open pits. Grades are retained in the block model for reconciling to life of mine information ●   Validation procedures are undertaken on the estimations. These include comparison of mean grades, visual comparisons to composite grades, slope of regression calculations, comparison to nearest neighbour analysis and Swath plots. As expected, Swath plots indicate increased smoothing outside of GC areas where exploration data is more limited. No conditional bias is observed.
Moisture	●   Tonnage estimates are completed on a dry basis. Both saprolite and transition samples are dried before SG determination.
Cut-off Parameters- Loulo
		Deposit		Yalea	Gara	Baboto	Loulo 3 (2015 costs)	Gara West (2015 Costs)
		Resource Type		UG	UG	OP	OP	OP
		Mining Cost	$/t milled	46.09	43.63	3.31	3.72	3.90
		Mine Sustaining Capital	$/t mined	7.67	8.25	-	-	-
		Processing Cost	$/t processed	17.80	18.20	15.20	20.60	20.60
		General & Admin Cost	$/t processed	7.80	7.70	7.80	8.80	8.80
		Crushing & Hauling Cost	$/t processed	-	-	3.63	-	-
		Royalty	%	6	6	6	6	6
		Gold Selling Price	$/oz	1500	1500	1500	1500	1500
		Processing Recovery	%	84.50	94.20	93.00	92.00	89.00
		Dilution	%	11.5	15.6	10.0	10.0	10.0
		Ore Loss	%	2	4	3	3	3
		Au Cut-off Grade (Marginal Ore)	g/t	2.04	1.89	0.73	0.73	0.75

18^th September 2018 Page 332

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

  Criteria     Commentary  

Cut-off Parameters - Gounkoto
	Deposit		Faraba		Gounkoto
	Resource Type		OP		OP		UG
	Mining Cost	$/t milled		3.52		2.95		49.3
	Mine Sustaining Capital	$/t mined				-		8.14
	Processing Cost	$/t processed		20.6		19.0		17.8
	General & Admin Cost	$/t processed		8.8		7.7		7.8
	Crushing & Hauling Cost	$/t processed		5.9		5.89		-
	Royalty	%		6		6		6
	Gold Selling Price	$/oz		1500		1500		1500
	Processing Recovery	%		91		92		94.2
	Dilution	%		10.0		10.0		13.4
	Ore Loss	%		3		3		4
	AuCut-off Grade (Marginal Ore)	g/t		0.8		0.80		2.3

Mining Factors or Assumptions - Loulo	●   In the open pit, the mining method assumed is conventional drill and blast followed by truck and excavator load and haul Fresh rock and transitional zones are drilled and blasted in 10 m lifts, with excavation of benches containing ore in three flitches. A small volume of weathered oxide material is free dig. This bench level concurs with the block heights set during the resource estimation to ensure minimal dilution.
	●   The Yalea and Gara underground mines are accessed via portals located in open pits and a box cut. The lower part of the mines has been developed as single declines with truck haulage up to crushers which feed ore and waste onto conveyors
	●   Three mining methods are proposed to be used at Yalea and Gara underground mines. They are long hole transverse open stoping; long hole longitudinal retreat open stoping; and stoping under rock fill (SURF).
	●   Longitudinal stoping in the upper areas of the mines is used in a panel and pillar method, where ~10 m long pillars are left between 50 m long panels. Paste fill plants at each of the mines produce a mixed paste / aggregate fill. With the introduction of paste fill the lower areas of the mines are being mined using an underhand (top down) long hole stoping method which retreats to central accesses in an echelon format. Panels are 50 m long, mined as single level stopes (25 m high) and filled with cemented paste fill. The paste fill is exposed by the mining of both the panel below and the next panel on the same level.
	●   Transverse long hole open stoping is used in the wider (>15 m wide) parts of Yalea North. Transverse stopes are 15 m to 30 m along strike and mined from hanging wall to footwall.
	●   The stoping under rock fill method (SURF) is proposed for the part of Yalea south upper that has weathered transition or saprolite present in the mineralised zone, or the immediate hanging wall. The method will be used as it is expected to provide a more consistent production in the poorer ground.
	●   Level accesses are at a vertical interval 25 m in all new long hole stoping areas; with 20 m level intervals in stoping under rock fill (SURF) areas and some upper areas of the mines which were developed with 20 m level interval are still in production.

18^th September 2018 Page 333

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   Ore loss in the 2017 Ore Reserve estimate has been included as 2% for Yalea stopes with fresh rock hanging wall and 4% for Gara stopes with fresh rock hanging wall.
	●   The economic and mining factors are well established at the Project
Mining Factors or Assumptions - Gounkoto	●   The Gounkoto open pit mine has been in operation since 2011, with an additional Super Pit feasibility study being undertaken during 2016. Therefore, the economic and mining factors are well established at the Project. Current Load and Haul operations occur on 10 m benches with flitch mining on three 3.3 msub-benches for levels containing ore. This bench level concurs with the block heights set during the resource to ensure minimal dilution. A 10% dilution and 3% ore loss factor is currently applied to reserves and a good 101% reconciliation between GC call and crusher tonnage and 101% on grade giving 101% reconciliation on total ounces reported for the year with short interval variability observed.
Metallurgical Factors or Assumptions	●   The Loulo processing plant uses a carbon in leach (CIL) gold extraction process with a throughput capacity of 4.8 Mtpa.
	●   The operating philosophy is to maintain process plant gold recovery above 90% by blending the various ore sources to control copper and arsenic grades in the mill feed. This blending process is anticipated to adequately manage copper and arsenic issues over the life of mine.
	●   Forecasted gold processing recovery is based on both testwork and operational history. The Yalea and Gara metallurgical recoveries for remaining LOM have been estimated at 84.5% and 94.5% respectively.
	●   The Gounkoto Super Pit testwork and historical operations data has indicated an estimated recovery of 92.5%.
	●   Yalea recovery is impacted by the presence of arsenic and copper which reduce recovery by impacting the process of adsorption of the gold onto the CIL tanks. Arsenic and copper impurities also increase cyanide and oxygen consumption. Consequently, arsenic and copper estimations are completed as part of the Mineral Resource update in order to identify potentially low recovery areas. Gold recovery is maintained above 90% by blending the various ore sources (Yalea / Gara/ Gounkoto) to control copper and arsenic grades in the mill feed.
	●   The current LOM has an average recovery life of mine of 92.3%. The average gold recovery in 2017 was 92.7%, an improvement from 2016 (Figure 1 2).
Environmental Factors or Assumptions	●   An environmental management plan (EMP) is in place, the Loulo operations are ISO 14001 compliant and independently audited to continuously improve environmental management. All environmental permits are in place for the Loulo processing plant and the Yalea / Gara underground mines and the Gounkoto Super Pit. The site is also audited against the requirements of the International Cyanide Management Code.
	●   Waste rock is generated from the open pit and underground operations; the waste rock dump at Gounkoto is close to the Falémé River which forms the border with Senegal. The waste dump has been carefully designed to minimise seepage and a collection system is in place. Waste rock will also be disposed of into the southern portion of the Gounkoto Super Pit to minimise surface disposal.

18^th September 2018 Page 334

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   Waste rock geochemistry has been studied and revealed potential acid rock drainage (ARD) generation and metals leaching from specific lithologies. However, this is closely monitored and no significant ARD issues have occurred.
	●   Tailings are generated from the Loulo plant and disposed of in the tailings storage facility (TSF) some 8 km east of the plant. The relevant Environmental and Social Impact Assessment (ESIA) and Management Plan for permitting requirements are underway for the expansion of the current TSF.
Bulk Density	●   Density measurements were carried out both within the mineralised domains and within defined waste units. Fresh, transitional, and saprolite material was obtained from the dry diamond drilling core and the water immersion method was used to calculate specific density. Density, for each mineralised domain, waste rock lithology and weathering zone was applied to the appropriate portion of each zone, except for within Yalea ‘Purple Patch’ where a single pass of inverse distance squared was used to interpolate density due to differing density populations. Any blocks not interpolated in a single pass were coded with a default density.
Classification - Loulo	●   Resource classification is clearly defined for the deposits. The classification criteria is based on the table below. Mineral Resources are classified as Measured, Indicated and Inferred Resources based on drilling density, geological continuity, and confidence, the variogram range continuity and the slope of regression.

Class	Criteria	Yalea	Gara	Loulo 3	Baboto	Gara West
Measured	Maximum Drill Spacing (m)	30	30	5 by 10	5 by 10	N/A
	Geological Continuity	6 Sections	6 Sections	6 Sections	6 Sections
	Min Samples	8	8	8	8
	Minimum Slope of Regression	0.8	0.8	0.8	0.8
Indicated	Maximum Drill Spacing (m)	30 to 80	40 to 80	20 to 80	20 to 60	10 to 40
	Sections of Geological Continuity	Good	Good	Good	Good	Good
	Min Samples	6	6	6	6	6
	Minimum Slope of Regression	0.6	0.6	0.6	0.6	0.6
Inferred	Maximum Drill Spacing (m)	Over 80	Over 80	Over 80	Over 80	Over 40
	Sections of Geological Continuity	-	-	-	-	-
	Min Samples	4	4	4	4	4

	●   Slope of regression plots and kriging efficiencies were also used to determine classification
	●   Classification shapes were generated using the above criteria and used to flag the block model to improve continuity. Blocks that did not fall within these criteria were flagged as unclassified.
Classification - Gounkoto	●   Resource classification is clearly defined for the deposits. The classification criteria is based on the table below. Mineral Resources are classified as Measured, Indicated and Inferred Resources based on drilling density, geological continuity, and confidence, the variogram range continuity and the slope of regression.

18^th September 2018 Page 335

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

  Criteria     Commentary  

Class	Criteria	Gounkoto Open Pit		Gounkoto Underground		Faraba
Measured	Maximum Drill Spacing (m)		12.5 by 12.5 (6 by 6 Finger Zone)		N/A		12.5 by 12.5
	Geological Continuity
	Min Samples
	Minimum Slope of Regression		0.7
Indicated	Maximum Drill Spacing (m)		30		40 by 30		30
	Sections of Geological Continuity
	Min Samples
	Minimum Slope of Regression
Inferred	Maximum Drill Spacing (m)		<100		<100		<100
	Sections of Geological Continuity
	Min Samples

	●   Slope of regression plots and kriging efficiencies were also used to determine classification
	●   Classification shapes were generated using the above criteria and used to flag the block model to improve continuity. Blocks that did not fall within these criteria were flagged as unclassified.
Audits or Reviews -Loulo	●   An external audit by QG reviewed the Yalea and Gara estimates in January 2015
	●   QG determined that the Yalea and Gara Mineral Resource estimates were “free of material errors”, and there were no issues that required immediate attention.
	●   QG identified that their primary minor concern was that the extents of the Indicated/Inferred boundary would require further drilling, which has since been completed.
	●   QG identified other incremental improvements, all of which have been implemented or addressed.
	●   QG’s analysis of the resource estimation through review of the wireframes, block model and macros suggested that there are no significant concerns in the resource estimation procedures. Their findings agreed with the use of different orientation domains to manage varying estimation orientations.
	●   QG stated that they consider the resource classification complies with the guidelines set out in JORC (2012) Code.
Audits or Reviews - Gounkoto	●   An external audit by QG reviewed the Gounkoto estimate in February 2015
	●   QG stated that the Gounkoto Mineral Resource was “free of error”.
	●   QG provided a number of incremental improvements, all of which have been implemented or addressed
	●   QG’s analysis of the resource estimation through review of the wireframes, block model and macros suggested that there are no material errors in the resource estimation procedures. Their findings agreed with the use of nine orientation domains to manage varying estimation orientations.
	●   The report identified that the greatest risk was the complex geological continuity. QG noted that the geological uncertainty could lead to

18^th September 2018 Page 336

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	higherore-loss and dilution if the deposit is mined underground. This risk is now been addressed through the completion of a pre- feasibility study on the underground material in 2016. In addition, the life of mine open pit has been deepened to include some formerly underground material.
	●   QG stated that they consider the resource classification complies with the guidelines set out in JORC (2012).
Discussion of RelativeAccuracy/ Confidence	●   The resource estimates have followed industry standard practices for collecting, validating, and estimating data.
	●   The application of thorough exploratory data analysis and quantitative neighbourhood kriging analysis gives a high confidence in the Measured and Indicated material in the model. The Inferred material by nature has a relative low level of accuracy, but a high geological confidence to exist.
	●   As Loulo and Gounkoto are operating mines, grade control and advanced grade control drilling have provided a much higher degree of accuracy in the model and the ability to reconcile mine data to compare with the block model
	●   Independent external audits are undertaken every three years, or at material model changes.to ensure that the procedures are appropriate.

18^th September 2018 Page 337

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Section 4. Estimation and Reporting of Ore Reserves

(Relevant criteria listed in section 1, 2, and 3 also apply to this section.)

Criteria	Commentary
Mineral Resource Estimate for Conversion to Ore Reserves	●   The Ore Reserve statement is based upon the Mineral Resource declared as of 31st December 2017 by Randgold Resources Ltd. Mineral Resources are reported inclusive of Ore Reserves. ●   The Yalea Ore Reserves are based on a geological block model produced in October 2017, Gara in November 2017, Gounkoto in July 2017, Baboto in May 2017, Loulo 3 and Gara West in November 2015. ●   Mineral Resources are supported by a dedicated Competent Person Report
Site Visits	●   Mr Derek Holm FSAIMM is a Qualified Person and an independent consultant with over 18 years of experience within the mining industry in technical, management and consulting positions. He visited the site between the 12thand 14thSeptember 2018
	●   The study to convert Mineral Resources to Ore Reserves is an operational life of mine plan update. The Yalea and Gara underground mines have an operating production history of 10 years at Yalea (9.9 Mt for 1.9 Moz) and 5 years at Gara (6 Mt for 0.8 Moz) respectively. The Competent Person has reviewed studies and operational history that support all material Modifying Factors and considers it is at least equivalent to Pre-Feasibility Study level.
	●   Mining of the Baboto pit is due to start in February 2018 with a small oxide pit. Mining will then cease and commence again in 2027, where the remaining transitional and sulphide ore will be mined.
Study Status	●   The Gounkoto Underground Ore Reserve is predicated upon a Pre-feasibility Study commissioned by Randgold Resources Ltd and completed in December 2017. The finding of the Feasibility Study was an economically viable mining operation that demonstrated lower operation risk, higher operational flexibility, and better overall cash flow to the smaller open pit and larger underground mine that formed the basis of the 2017 LOM plan.

18^th September 2018 Page 338

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	Loulo Open Pits:
		Parameter	Unit	Gara West	Baboto	Loulo 3
		Gold Price	$/oz	1,000
		Royalty	%	6%
		Selling cost	%	0%
		Net Gold Price	$/oz	940
		Met Recovery	%	91%
		Dilution	%	10%
		Ore Loss	%	2%
		Mining Cost - Contractor	$/t mined	3.63	3.18	3.45
		Mining Cost - Owner’s	$/t mined	0.06	0.06	0.06
		Mining Cost - Grade Control	$/t mined	0.21	0.07	0.21
		Total Mining Cost	$/t mined	3.90	3.31	3.72
		Strip Ratio	Waste/Ore	1.94	4.5	1.94
		G&A	$/t milled	8.80	7.80	8.80
		Ore Crushing & Hauling	$/t milled	0	3.63	0
		Mining	$/t milled	11.46	18.18	10.93
Cut-Off Parameters - Loulo		Process Plant	$/t milled	20.6	15.2	20.6
		Total Operating Costs	$	40.86	44.81	40.33
		Full Grade Cut-off	g/t	1.52	1.62	1.50
		Marginal Cut-off Grade	g/t	1.09	0.96	1.09
		Diluted Full Grade Cut-off Grades	g/t	1.38	1.48	1.36
		Diluted Marginal Cut-off Grades	g/t	0.99	0.88	0.99
	●   The full grade cut-off is the ore material that is profitable at $1,000/oz, considering all operating costs of mining, haulage, processing, and general and administrative costs, as well as the appropriate recovery, dilution, and realised gold price post Royalty at $1,000/oz spot gold price . It is the principal material fed to the plant.
	●   The marginal cut-off grade is the ore material mined within the $1,000/oz pit design that is profitable on a marginal basis (FG OPEX less mining costs).

18^th September 2018 Page 339

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	Loulo Underground
		Parameter	Unit	Yalea	Gara
		Gold Price	$/oz	1,000	1,000
		Royalty	%	6%	6%
		Selling cost	%	0%	0%
		Net Gold Price	$/oz	940	940
		Met Recovery	%	84.47%	94.2%
		Dilution	%	11.5%	15.6%
		Ore Loss	%	2%	4%
		Mine OPEX Development	$/t milled	5.00	3.29
		Mine Stoping	$/t milled	12.15	11.79
		Backfill	$/t milled	11.58	10.51
		Fixed Cost	$/t milled	13.98	14.63
		Grade Control	$/t milled	3.37	3.40
		Mine Sustaining Capital	$/t mined	7.69	8.25
		Total Mining Cost	$/t mined	53.78	53.78
		G&A	$/t milled	7.80	7.70
		Process Plant	$/t milled	17.80	18.20
		Total Operating Costs	$	79.38	79.68
		Breakeven In-Situ Cut-off  Grade	g/t	3.54	3.29
		Breakeven Mined Cut-off Grade	g/t	3.11	2.73
		Marginal In-situ Cut-off Grade	g/t	2.97	2.80
		Marginal Mined Cut-off Grade	g/t	2.61	2.33
	●   The cut-off grade values are calculated on the basis of mined gold grade, including dilution, in grammes per tonne. No by product credits or metal equivalents are used. Revenue calculation is based on a gold price in $ and takes into account processing losses. Costs are based on 2017 actual operational costs in $. The cut-off grades used are Yalea 2.4 g/t Au and Gara 2.3 g/t Au.

18^th September 2018 Page 340

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	Gounkoto Open Pit
	●   In the open pit, the weighted average cut-off grades for the Gounkoto and Faraba open pits are 1.20 g/t Au and 1.32 g/t Au respectively. Table 15-18 shows the economic parameters used for estimating the cut-off grade for the Gounkoto open pit and Faraba open pit, in determining the Ore Reserves.
		Parameter	Unit	Gounkoto	Faraba
		Gold Price	$/oz	1,000
		Royalty	%	6%
		Selling cost	%	0%
		Net Gold Price	$/oz	940
		Met Recovery	%	91%
		Dilution	%	10%
Cut-Off Parameters - Gounkoto		Ore Loss	%	2%
		Mining Cost - Contractor	$/t mined	2.95	3.45
		Mining Cost - Owner’s	$/t mined	0.06	0.06
		Mining Cost - Grade Control	$/t mined	0.07	0.14
		Total Mining Cost	$/t mined	3.08	3.65
		Strip Ratio	Waste/Ore	13.62	3.61
		G&A	$/t milled	7.70	8.80
		Ore Crushing & Hauling	$/t milled	5.89	5.9
		Mining	$/t milled	44.98	16.81
		Process Plant	$/t milled	19.0	20.6
		Total Operating Costs	$	77.57	52.11
		Full Grade Cut-off	g/t	2.85	1.95
		Marginal Cut-off Grade	g/t	1.20	1.32
		Diluted Cut-off Grades	g/t	2.59	1.77
		Diluted Marginal Cut-Off Grades	g/t	1.09	1.20

18^th September 2018 Page 341

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	Gounkoto Underground
	●   Underground mine operating costs were determined using November 2012 contractor mining costs, which formed a basis for the December 2016 cut-off grade for the Ore Reserve estimate. An insitu cut-off grade of 3.0 g/t Au was used for the stope design in the main part of the orebody, and an insitu cut-off grade of 4.0 g/t Au was used for the stope design in more isolated areas of the orebody. No supporting calculation was provided for these cut-off grades. As a check, a new and current cut-off grade was determined, resulting in a marginal insitu cut-off grade of 3.34 g/t Au. This is approximately in line with the combined 3.0 g/t Au and 4.0 g/t Au cut-off grade and compares favourably to the marginal insitu cut-off grades of Yalea and Gara of 2.97 g/t Au and 2.80 g/t Au, which are similar operations. The Gounkoto underground ore will be toll treated at the Loulo mill. The total processing cost applied is based on the current costs for the Loulo mill. The inputs to the revised cut-off grade calculations are shown in Table 15-20.
		Parameter	Unit	Gounkoto
		Gold Price	$/oz	1,000
		Royalty	%	6%
		Selling cost	%	0%
		Net Gold Price	$/oz	940
		Met Recovery	%	92.0%
		Dilution	%	710.2
		Ore Loss	%	6.3%
		Mine Capital Development	$/t mined	34.36
		Total Mining Cost	$/t mined	49.52
		G&A	$/t milled	9.80
		Process Plant	$/t milled	23.60
		Total Operating Costs	$	117.28
		Breakeven In-Situ Cut-off Grade	g/t	4.96
		Breakeven Mined Cut-off Grade	g/t	4.22
		Marginal In-situ Cut-off Grade	g/t	3.34
		Marginal Mined Cut-off Grade	g/t	2.84
	●   Optimisation of the cut-off grade strategy for Gounkoto will be conducted in the feasibility study.

18^th September 2018 Page 342

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
Mining Factors or Assumptions	●   The Ore Reserves have been estimated using three dimensional mine designs. These have been created using three dimensional geological wireframes and take into account the geotechnical environment.
	●   Open pit Ore Reserves use a dilution figure of 10% and an ore loss of 2% are based on historical results, which has Proved to be suitable based on the orebody and mining fleet equipment. Dilution is applied to both tonne and grade (increases the tonnes and drops the grade) in the optimisation and mining schedules. It simulates the waste rock that is included into the ore during the mining operation.
	●   The predominant UG mining method used is long hole open stoping (bench stoping and transverse long hole stoping). These methods are currently used at Yalea and Gara. Mining factors are supported by reconciliation of previously mined areas. Stope sizes (up to 25 m vertical level interval and up to 50 m panel length) are based on geotechnical analysis of stable spans. A stoping under rock fill (SURF) mining method is proposed to be used in the Yalea south upper area where weathered oxide and transition rock is present in the ore zone and immediate hanging wall.
	●   Planned dilution (within stope shapes) has been added where the geological boundary flexes between levels, or where ore drives have been placed in the footwall to avoid weathered zones. Planned dilution is estimated as 3% at Yalea and 5% at Gara.
	●   Unplanned dilution (outside stope shapes) has been added based on an equivalent width added to the hanging wall of the stope. Thicknesses were determined from reconciliation of previously mined areas . Unplanned dilution is estimated as 9% for Yalea long hole stoping and 11% for Gara long hole stoping. For the stoping under rock fill areas 35% dilution has been added. Dilution has been added at a grade of 0 g/t Au.
	●   In long hole stoping areas (fresh rock ore and hanging wall) a mining recovery factor of 96% has been used . For Stoping under rock fill areas, a mining recovery is applied based on ore rock type (oxide - 50%; transition-65%; fresh - 80%)
	●   The Loulo life of mine plan does not include any Inferred Mineral Resources at Gara or Yalea underground mines.
	●   The majority of the infrastructure required for the mining is already in place. The life of mine plan includes provision of additional infrastructure for mine ventilation, refrigeration, and dewatering.
	●   The Gounkoto Underground mining method is longhole stoping and longhole with backfill and incorporates both transverse and longitudinal access as appropriate to the width of the orebody. The minimum mining width for stopes is 3 m. Ground conditions have been estimated to be in the majority ‘Fair’, with minor inclusions of ‘Poor’ ground. Stope designs have been based on a stope stability assessment for stable stope surfaces. Therefore, unplanned dilution is expected to be low provided good mining practice is followed.
	●   Transverse stopes will be mined in a primary/secondary stope sequence. Cemented fill will be used to maximise ore recovery.
	●   Stopes have been designed for longhole stoping using a “top hammer” blast hole drill. The complexity of the orebody requires that some waste and Inferred and unclassified material may be included in stope designs so that suitable recovery of the mineralisation above cut-off grade is achieved. This mineralised waste, Inferred and unclassified material is classed as planned dilution. None of the proposed stopes report below 3 g/t Au. The proportion of inferred and unclassified material that is included in stope designs is <1.0% of the total stope tonnes. This is considered to be acceptable.
	●   Dilution and recovery factors have been modified where blind uphole stopes are used because of the reduced flexibility in blast hole drilling geometry.

18^th September 2018 Page 343

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   In regard to transverse primary stopes, waste dilution only occurs on the H/W contact.
	●   Dilution for transverse primary stopes has been set at 5%. Dilution for transverse primary blind uphole stopes has been set at 7.5% to take account of the more difficult mining conditions.
	●   In regard to transverse secondary stopes, waste dilution can occur from the H/W contact and the cemented fill.
	●   Dilution for transverse secondary stopes has been set at 7.5%.
	●   Dilution for transverse secondary blind uphole stopes has been set at 10%. Recovery for conventional transverse stopes has been set at 95%. Recovery for transverse blind uphole stopes has been reduced to 92%.
	●   In regard to the longitudinal stopes the dilution factor has been set at 10%. Recovery for conventional longitudinal stopes has been set at 95%.
	●   Recovery for blind uphole longitudinal stopes has been set at 79% which accounts for 5 m rib pillars being left every 25 m and 95% recovery of the remaining stopes. Recovery for longitudinal blind uphole stopes is better than transverse blind uphole stopes because of better blast hole drilling geometry in narrower stopes.
	●   Comprehensive 3D numerical modelling was completed on the stope designs and sequence. The results of that numerical modelling concur with the stability assessment.
	●   The underground mine design includes infrastructure suitable for the planned mining method and production rate using an access decline from the existing open pit.
Metallurgical Factors or Assumptions	●   Recovery in the CIL plant is based on suitable testwork and historical reconciliations with the metallurgical plant. The recovery applied to the Loulo open pit reserve is 91% for Gara West and Loulo 3 and 93% for Baboto. The recovery is estimated to be 92% and 91% for Gounkoto Open pit Ore and Faraba Open pit Ore respectively.
	●   The ore is to be processed through the existing Loulo CIL process plant. The process is well tested and appropriate for style of mineralisation. The process plant has been treating Yalea and Gara ore since 2005 and has demonstrated consistent recovery performance.
	●   Separate metallurgical domains have been created for oxide, transition, and fresh ores in each of the deposits. All underground diamond drill grade control samples are assayed for copper. Copper grades are interpolated in geological block model to assist with recovery prediction.
	●   The Yalea deposit contains higher levels of arsenic and copper than Gara. Historically these have impacted gold recovery, particularly in weathered material from Yalea open pit (South Area). The current strategy of feeding the process plant with a blend of Yalea, Gara and Gounkoto ores has been successful in achieving overall recoveries in excess of 90%. This strategy is anticipated to be used throughout the life of mine.
	●   Overall the Gounkoto ore responds well to direct cyanidation with an average of 92% recovery achieved in previous testwork programmes, Good reconciliation is being achieved from ore currently mined ore from the Gounkoto open pit which is located immediately above the proposed underground operation.
	●   It was recommended to use a flat gold recovery of 92% for the Gounkoto UG pre-feasibility study.

18^th September 2018 Page 344

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
Environmental	●   All ESIAs and Environmental Management Plans for the Loulo and Gounkoto operations are in place; the environmental permits, renewable every five years, were approved in March 2017 and the operation is ISO 14001 certified.
Infrastructure	●   Substantial infrastructure already exists at the Loulo site as part of the previous open pit mining operation. Ore processing and tailings facilities already exist at Randgold’s Loulo operation.
	●   The Life of Mine Plan includes provision of additional infrastructure for mine ventilation, refrigeration, and dewatering.
	●   Substantial infrastructure already exists at the Gounkoto site as part of the existing open pit mining operation.
	●   Surface infrastructure associated with Gounkoto Open pit operation includes the water diversion trench, camp, power station, surface workshops, offices and all other associated infrastructure required to support the open pit. All of these items have been designed, costed, and accounted for in the economic assessment of the Project.
	●   Minor additional surface infrastructure associated with the underground Gounkoto operation is required to the camp, power supply, surface workshops, offices and all other associated infrastructure required to support the u/g project. All of these items have been designed, costed, and accounted for in the economic assessment of the Project to Pre-Feasibility level.
Costs	●   Loulo operating costs (mining, processing, site general and administration) are derived from 2017 actual costs at Loulo operation (including Yalea and Gara underground mines). Life of mine costs have been adjusted to allow for operational changes including changes in paste fill binder.
	●   Development capital costs are derived from mine designs and are costed by allocation of a portion of the total development cost during each period. The allocation cost includes a share of mining overhead costs. Plant and equipment capital costs have been estimated by year for the life of mine.
	●   Nil allowances were made for deleterious element as there are no significant deleterious elements are present in the gold bullion.
	●   Transportation of gold bullion from site and refining charges are included in operating costs for the processing plant and are derived from existing agreements.
	●   A royalty of 6% of gold sales is payable to the Malian government.
	●   Gounkoto mining costs are based on contract mining costs agreed with the Mining Contractor GMS and the Gounkoto Mine. Owner mining costs of grade control, planning, dewatering and included in the budget numbers.
	●   The costs for processing and G&A were updated based on operational data adjusted to forecast production profiles and manning levels for Loulo-Gounkoto.
	●   For the Gounkoto Underground Pre-Feasibility Study, the Loulo Operation 2016 mine budget rates were applied where relevant. These are based on an owner mining approach using expat’ labour to conduct the development and production of the mine as well as training of a national workforce over time.
	●   In 2015, Feasibility level budget costs were obtained for a larger scale underground operation for infrastructure items such as ventilation fans, refrigeration plant, CAF and CRF plant equipment and pumps. These costs were reviewed and applied into the Pre-Feasibility cost model.
	●   Costs for processing and G&A were applied based on the 2016 production profiles and manning levels for Loulo-Gounkoto.

18^th September 2018 Page 345

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   Customs duties, taxes, charges, and logistical costs are included in all relevant areas of the operational costs.
Revenue Factors	●   A gold price of $1,000/oz was employed for the Ore Reserve process. This value is based on a conservative view of gold price following a period of relatively higher gold price volatility.
Market Assessment	●   The gold market is highly liquid and benefits from terminal markets (London, New York, Tokyo, and Hong Kong) on almost a continuous basis. Gold prices were in general on a downward trend from 1980 to 2000 where it traded down to approximately $250/oz. Between 2000 and 2011 the market was on a general upward trend that moved spot prices to a peak of $1,900/oz. 2013 saw a sharp correction in the upward trend, with the spot price dropping to $1250/oz. Since 2014 the gold price, has traded in range of $1,050 to $1,400 / troy. Oz. Gold is currently (August 2018) trading at approximately $1,200/oz.
	●   Gold produced at the mine site is shipped from site, under secure conditions, to a refining company. Under pre-established contractual conditions, the refiner purchases the gold from the mine with the proceeds automatically credited to the mines’ bank account. The operation is unhedged.’
Economic	●   The economic analysis of the life of mine uses a range of discount rates from 2.5 to 10%. No inflation of costs or revenues has been included in economic analysis. Sensitivity analysis to gold price, mined grade and operating costs have been undertaken. At Yalea the economic outcome is most sensitive to gold price. At Gara the economic outcome is highly sensitive to all of these factors due to the lower mined grade and profit margin.
	●   LOM Capital: $598M including construction, ongoing, capital development, waste stripping, drilling, pre-production, exploration, rehab and closure.
Social	●   The mine has not faced any material social issues. However, there is a growing need to reinforce communication and sensitization in the community regarding mining and its benefits. The mine will continue its engagement to manage this potential risk.
	●   The various ESIAs carried out for Loulo and Gounkoto have included assessments of impacts and benefits to local communities. A comparative socio-economic survey has been conducted to assess the change (positive and negative) in the surrounding local communities over an eighteen-year period (1997 to 2015). Since 2002 there has been a significant influx of people into the area of the mine. Modernization of the communities has increased along with trade and commerce. Educational and skill levels have also increased but access to land to grow crops or graze livestock, and access to natural resources have decreased with higher populations. The quality of the housing stock has increased, along with transport infrastructure, access to water, access to healthcare and education.
	●   Stakeholder engagement and dialogue is ongoing and the importance of this to the company. In addition, Union representatives attend strategy meetings and quarterly review meeting in order to provide information on current community and local employee issues.
	●   A grievance mechanism is in place. No grievances were received in 2017; the last grievance was recorded in 2015.
	●   The mine is a significant employer to members of the local communities. The underground mining operations contribute to extended life-of-

18^th September 2018 Page 346

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	mine, employment of local Malians and the growth of the Malian economy. Randgold’s policy is to promote nationals to manage the project.
	●   Randgold’s policy of promoting local employment also extends to its contractors. Figures for national employment in 2017 were 2,820 out of a total of 2,975 employer and contractor posts. Unskilled labour is typically sourced from the local area while more skilled posts are filled by staff from elsewhere in Mali, including Bamako and the other mines currently operating in the country.
	●   Randgold continually invests in the local communities and focuses on potable water supplies, primary school education, health care and education.
	●   A program to improve the agricultural yield of the area has been undertaken, with tractors being donated to the community and an agribusiness training centre constructed to train locals in agricultural entrepreneurship; and to develop community vegetable growers to supply the mine caterer with fresh produce.
	●   Phases of economic displacement (loss of crops and trees) and the physical resettlement of some households has occurred during the life of the mine to date as it has expanded. Crops and trees are compensated at published rates; physical resettlement has involved providing affected households with a new home. The most recent resettlement was around Gounkoto and was completed in 2012. This affected 12 households and around 300 hectares of land, including 1,700 economic trees and two artisanal mining (orpaillage) sites. A resettlement audit was carried out in late 2015, in line with good international industry practice.
	●   There is a significant ongoing presence of artisanal miners (orpailleurs) operating within the Loulo Permit area, particularly along the haul road. The government of Mali, as part owners of the operations, provide security in the form of Gendarmes who intervene on request if ASM activities interfere with the operation of the mine. The Gendarmes have been successful in removing orpailleurs from the Baboto target area. Although this and other key exploration targets remain free of ASM activity, the risk of incursion into exploration or operations areas remains, because of the increasing number of people in the largely illegal (i.e. unlicensed) activity. The mining industry in Mali has established a committee to deal with the issue at the highest level of the Government. Dedicated orpaillage corridors are still to be created for artisanal and small- scale mining. The World Bank has recently been involved in finding solutions. In the meantime, the Mine is reinforcing its relationship with the community to deal with the issue.
Other	●   All necessary governmental agreements and approvals critical to the viability of the Project are in place.
	●   Loulo operates within the Loulo Permit which authorises the carrying out of exploration and mining activities. The Loulo permit extends beyond the current Yalea and Gara life of mine and is renewable.
	●   The most significant risks in the area are the growth in numbers of illegal miners, which has seen a rapid expansion since the improvement in infrastructure between Dakar and Bamako.
	●   The Gara mine is within 500 m of the Falémé River, it does have 1 in 100 year flood protection, however the risk remains of extraordinary flooding causing delay in operations.
	●   All necessary governmental agreements and approvals critical to the viability of the Project are in place.

18^th September 2018 Page 347

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
Classification	●   The Loulo Ore Reserves are classified as Proved and Probable based upon the confidence levels determined in the Mineral Resource and accurately reflects the Competent Person’s views of the deposit.
	●   Ore Reserves underground at Yalea and Gara are classified as Proved and Probable Ore Reserves.
	●   Ore stockpiles on the surface are classified as Measured Mineral Resource and Proved Ore Reserve. These stockpiles are reported as part of the Loulo Open Pit Ore Reserves estimate.
	●   The Loulo and Gounkoto Open Pit Ore Reserves are classified as Probable based upon the confidence levels determined in the Mineral Resource and accurately reflects the Competent Person’s views of the deposit.
Audits or Reviews	●   An external audit of the Mineral Resource was completed in February 2015 by QG. This audit stated that QG is confident that the Loulo Mineral Resources are free of material error. There are no issues that QG consider need immediate attention. Any recommendations made are more for long term improvement in processes and understanding
	●   QG’s analysis of the resource estimation through review of the wireframes, block model and macros suggested that there are no material errors in the resource estimation procedures. Their findings agreed with the use of different orientation domains to manage varying estimation orientations.
	●   QG stated that they consider the resource classification complies with the guidelines set out in JORC (2012).
	●   All outcomes of the external audits have been acted upon where relevant.
	●   During 2013 -2016 the Yalea and Gara Life of Mine Planning was reviewed by Kenmore Mine Consulting. Kenmore has concluded that the current mine planning was robust but recommended further work to improve confidence in mine ventilation, paste fill testwork, geotechnical rock mass modelling, pre- mining stress measurement and mining induced stress modelling. Randgold has undertaken substantial work to address these recommendations and this work is ongoing as part of a continuous improvement program.
	●   Additional external expert opinion was sought to provide clarity and finality on decisions such as ventilation and refrigeration requirements and ground stability issues. The mine design has been reviewed by peers and senior Randgold management from the sister operation at Loulo. All items considered of a critical nature by the formal reviews and external experts were resolved and considered in the Ore Reserve.
	●   An external audit on the Gounkoto Mineral Resource and input data procedures was completed in February 2015 by QG. This report stated that sampling procedures for RC and diamond drilling was good. Some minor improvements for the collection bulk density measurements (use of recording balance and certified weights) were suggested and were implemented during 2016.
Discussion of Relative Accuracy/ Confidence	●   The modifying factors used are considered realistic for the Project based on the mining methods selected and assuming good practise is applied. The values applied are in-line with published data from similar scale operations.
	●   Accuracy and confidence level in the Ore Reserve estimate has been assessed qualitatively.
	●   Yalea and Gara deposits have a consistent production history over the last 10 years both as open pit and underground mines. Reconciliation process have been undertaken over this time to drive continuous improvement in the operation.

18^th September 2018 Page 348

Loulo-Gounkoto Gold Mine Complex NI 43-101 Technical Report

Criteria	Commentary
	●   The key risks / opportunities that may effect the relative accuracy and confidence of the Ore Reserve are: Gara mined grade - due to the relatively lower economic margin at Gara; Gara dilution - due to narrow vein width and higher variability.
	●   The Gounkoto Open Pit Ore Reserve estimate is classified as Proved and Probable in-line with the Mineral Resource classification and modifying factors applied to the Mineral Resource.
	●   Due to the complex nature of the deposit it is recognised there will be local deviations to the Ore Reserve estimate hence the level of confidence is placed as Probable. However, on a global basis there is no evidence of significant risk to the Ore Reserve.
	●   The modifying factors used are considered realistic for this project based on the geotechnical environment and mining methods selected and assuming good practise is applied. The values applied are in-line with published data from similar scale operations.
	●   The reconciliation between the mined estimated tonnes and grade reconcile well with the weightometer and grade sampling with a +0.6% variance on tonnage, +0.7% variance on grade and 1.3% variance on ounces.
	●   The Gounkoto Underground Ore Reserve estimate is rated as probable in-line with the Mineral Resource classification. A 3rd party review of the Resource model was completed in 2015. The findings were satisfactory.
	●   Due to the complex nature of the deposit it is recognised there will be local deviations to the Ore Reserve estimate hence the level of confidence is placed as Probable. However, on a global basis there is no evidence to suggest significant risk to the Ore Reserve.
	●   The modifying factors used are considered realistic for this project based on the geotechnical environment and mining methods selected and assuming good practise is applied. The values applied are in-line with published data from similar scale operations.

18^th September 2018 Page 349