Bema Gold 6-K New Release Bema Announces Preliminary Mineral Resource for the Kupol Project

New Release
Bema Announces Preliminary Mineral Resource for the Kupol Project

Vancouver, February 3, 2004- Bema Gold Corporation (BGO-TSX, AMEX, BAU-AIM) is pleased to announce a preliminary Mineral Resource from the Kupol gold and silver property located in Chukotka, north eastern Russia. This resource estimate confirms that the Kupol vein system hosts a large, high grade gold and silver deposit. Bema has estimated the preliminary Mineral Resource based on 21,860 metres of drilling in 2003. Significant gold mineralization has been intercepted over 3.1 kilometres of drilled strike length and, from surface, to a depth of at least 300 metres. Mineralization remains open to the north, south and at depth.

The following tables show a risk adjusted Mineral Resource and an uncut Mineral Resource for the Kupol vein. In addition, the tables include, separately, the resource estimates for the Big Bend zone, the highest grade zone to date.

Mineral Resource, Risk Adjusted
Indicated Mineral Resource, 6 g/t gold cutoff

	Tonnes (000’s)	Gold (g/t)	Silver (g/t)	Gold Ounces (000’s)	Silver Ounces (000’s)
Big Bend zone	1,257	29.6	308	1,196	12,400
Kupol Vein, all zones	2,551	22.3	232	1,826	19,100

Inferred Mineral Resource, 6 g/t gold cutoff

	Tonnes (000’s)	Gold (g/t)	Silver (g/t)	Gold Ounces (000’s)	Silver Ounces (000’s)
Big Bend zone	2,629	22.4	337	1,895	28,500
Kupol Vein, all zones	7,166	18.4	243	4,231	55,900

The Mineral Resource is risk-adjusted with an average 11% metal reduction in high-grade Indicated, 13% in low grade Indicated and 4% metal reduction in the Inferred Resources. These risk-adjustments have been made to account for a large portion of the gold being represented by a relatively small proportion of samples having very high grades. The metal at risk is typical of that found for projects at the Prefeasibility Study stage. Typically during Feasibility (with additional data), this metal reduction would be approximately 5% for Indicated. The risk adjusted resource was estimated using kriging and has been prepared in accordance with National instrument 43-101 under the supervision of Bema’s Qualified Person Tom Garagan in consultation of Dr. Harry Parker of AMEC Americas.

An uncut resource was also prepared to reflect the potential in the Indicated and Inferred areas.

Mineral Resource, Uncut

Indicated Mineral Resource, 6 g/t gold cutoff

	Tonnes (000’s)	Gold (g/t)	Silver (g/t)	Gold Ounces (000’s)	Silver Ounces (000’s)
Big Bend zones	1,257	32.8	337	1,326	13,600
Kupol Vein, all zones	2,551	25.2	259	2,063	21,200

Inferred Mineral Resource, 6 g/t gold cutoff

	Tonnes (000’s)	Gold (g/t)	Silver (g/t)	Gold Ounces (000’s)	Silver Ounces (000’s)
Big Bend zone	2,629	23.1	355.0	1,954	30,000
Kupol Vein, all zones	7,160	19.1	250.0	4,387	57,600

1

Due to the wider spaced drilling in some portions of the Central and North Extension zones, significant sections of the Kupol vein are not included in these preliminary resource estimates and require closer spaced drilling to be included in the next resource calculation.Please refer to the long section map available at www.bema.com.

An infill and exploration program comprising 57,000 meters of diamond drilling with 7 drills will commence in late May. Infill drilling will begin at Big Bend and exploration drilling will commence with deep drilling in Big Bend, the North Zone Extension and stepout drilling to the north. Management believes that the ultimate target at Kupol is up to, or in excess of, 16 million tonnes with similar gold and silver grades to those stated in the preliminary mineral resource.*

Metallugical test work for Kupol commenced in late December 2003. Preliminary results from a series of bottle roll cyanidation tests indicate gold recoveries from 94.5% to 96.5% and silver recoveries ranging between 70% and 80%. Additional test work to determine optimum grind size and reagent mix is continuing. Furthermore, a preliminary economic assessment of the project will be completed in May, 2004.

Bema is currently procuring equipment and supplies to the Kupol project preparing for the 2004 exploration and development program. The program will include construction of a runway for fixed wing aircraft, earth works for mine and mill facilities, geotechnical and condemnation drill programs, final metallurgical test work, and procurement of equipment for 2005 construction.

The Kupol Project is a joint venture between Bema and the government of Chukotka. Bema has the option to purchase a 75% interest in the property from the government

On behalf of BEMA GOLD CORPORATION

“Clive T. Johnson”
Chairman, C.E.O., & President

For more information on Bema Gold please contact Investor Relations at (604) 681-8371 or toll-free 1-800-316-8855 or alternatively contact our web-site at www.bema.com

*The potential quantity and grade is conceptual in nature and there has been insufficient exploration to define this target at this time and it is uncertain that further exploration will result in further discoveries on the property. The target referred to is based on follow up to currently untested, known strike length and down dip potential.

The descriptions of methodology are below:

The resource was estimated from a 3 dimensional block mode, that was created by interpreting the vein, stockwork zone, dyke and faults on east-west trending cross sections and reconciling theinterpretations on levelsl. Three-dimensional solids (wireframe) models were built from the interpretations and were the basis for coding the block model. Assay intervals within the vein and stockwork were capped before compositing and block grade estimation. The high-grade capping was set at 300 g/t Au and 3200 g/t Ag for vein and 30 g/t Au and 300 g/t Ag for stockwork. This was further modified with a metal at risk adjustment. After kriging, the grades were modified using a metal at risk adjustment explained in detail at the end of the news release. The adjustment has removed a total of 11.2% of the metal from the uncapped high grade indicated resource, 13% from the low grade indicated and 3.6% from the inferred. This adjustment would be in the order of 5% for an indicated resource with infill drilling at feasibility.

2

Within the vein and stockwork interpretation, 1.5 meter composites were created from the capped assay intervals. To best represent the high and lower grade portions of the vein, an indicator variable was created using the intensity of sulfo-salts and Au grade as a guide. Variograms were run on composites for the indicator variable in addition to Au and Ag within the high-grade portion and Au and Ag within the lower grade portion of the vein.

The indicator (high-sulfosalt) variable, high-grade Au, low-grade Au, high-grade Ag and low-grade Ag were estimated with four passes of ordinary kriging. Each of the passes was used to control the amount of data mixing with consideration given to the drill hole spacing. Six different search ellipse domains for block grade estimation were placed from north-to-south to match the local strike and dip of the vein.

The whole block grade for vein was calculated using indicator kriging which involves weighting the high and low grade portions by the indicator. The formula used in this calculation was determined by visual inspection of the block grades relative to the drill hole and trench data and comparison of the average grade at a zero cutoff to the average of the declustered composites. A specific gravity of 2.55 was used for tonnage calculations.

Checks made on the model include a comparison of the kriged estimate to the nearest neighbor (block polygon) models, to the declustered composites, and to the distribution expected from a 10 x 3 x 5 meter selective mining unit. This selective mining unit accounts for internal dilution within the vein. The affect of edge (contact) dilution and ore loss will be assessed in a mining study that is in progress. Visual checks of the block grades relative to the drill hole and trench data was completed on cross sections and levels on the computer screen and on paper plots of the cross sections.

Au and Ag grades in stockwork were estimated using inverse distance to a power of eight, with a similar searching strategy used for estimating vein.

Indicated and Inferred resources were defined by reviewing grade and mineralized vein width on east/west trending cross sections and on a vertical longitudinal projection. Indicated resources are defined within continuously mineralized areas that are drilled to 50 metre spacing or less, with some of it on a 25-metre drill spacing. Projection of Indicated Resources is limited to 25 metres down-dip in the vein and 12.5 to 25 metres along strike. Most of the Indicated Resource occurs within the Big Bend area. Inferred resources are limited to 100-metre projection down-dip from the last vein intercept at depth. Minor local projection of Inferred resources is extended to 150 metres down-dip of drilling.

The Kupol QA/QC program used to monitor the accuracy of the assay database was managed by Bema’s Qualified Person Tom Garagan and was audited by Smee and Associates.who found it to be compliant with 43-101 regulations. The lithology database was verified by redundant checks against original and quick log information. Modeling was performed using GEMCOM software by AMEC Americas and Bema personnel, with the modeling process overseen by Susan Meister (Resource Modeling consultant) and Harry Parker (AMEC, Senior Resource Modeler).

Comments on Resource Categorization:

Mineral Resources have been categorized using the classification of the Canadian Institute of Mining, Metallurgy and Petroleum (2000), relevant definitions being quoted below. This classification is the basis for Technical Reports by Qualified Persons in Canada, and the classification is virtually the same as that of the JORC code (Australia), SME guidelines (USA), SAMREC (South Africa) and that of the European Union.

At Kupol, Indicated Mineral Resources are estimated where drill holes or trenches intersect the vein(s) at approximately a 50 m spacing on a vertical longitudinal projection. The Indicated Resources are generally located within 100 m of the surface, and the majority are located in the Big Bend and Central Sectors. With this spacing, the vein structure is continuous, although the vein thickness may be locally affected by

3

faulting and removed by dikes. The grade appears continuous from hole to hole, and this continuity has been confirmed by approximately 104 trenches spaced at 4 m intervals across the outcrop of the vein in the Big Bend and Central Sectors. The resource model estimates that 71 percent of the vein meets a cutoff grade of 6 g/t, with 84 percent of the vein meeting a cutoff grade of 6 g/t in the Big Bend Sector. Trenches show that the veins are often braided with low-grade or barren stockwork and gouge-filled shears. Infill drilling will be required to support feasibility studies and mining operations.

Inferred Mineral Resources are estimated down-dip and along strike from Indicated Resources in areas that have been drilled on approximately a 100 m spacing (locally 150 m spacing) on a vertical longitudinal projection. Preliminary geostatistical studies and inspection of sections show that continuity of grade is greatest in the down-dip direction. Projection distances have been limited to within 100 m of a drill hole, locally 150 m (vertically), with the depth of Inferred Resources extending to approximately 350 m from surface. There are four holes at this depth (KP046, KP107, KP134, KP144) that show the continued existence of the vein structure and mineralization. The current drill hole spacing is too wide to enable grade continuity to be demonstrated. The resource model assumes that 50 percent of the vein meets a cutoff grade of 6 g/t, with 80 percent of the vein meeting a cutoff grade of 6 g/t in the Big Bend Sector.

CIM Mineral Resource Definitions

An Indicated Mineral Resource is that part of a Mineral Resource for which quantity, grade or quality, densities, shapes and physical characteristics can be estimated with a level of confidence sufficient to allow appropriate application of technical and economic parameters, to support mine planning and evaluation of the economic viability of the deposit. The estimate is based on detailed and reliable exploration information gathered through appropriate techniques from locations such as outcrops, trenches, pits, workings and drill holes that are spaced close-enough for geological and grade continuity to be assumed.

An Inferred Mineral Resource can be estimated on the basis of geological evidence and limited sampling and reasonably assumed, but not verified geological and grade continuity. The estimate is based on limited information from locations such as outcrops, trenches, pits, workings and drill holes.

Due to the uncertainty which may attach to Inferred Mineral Resources, it cannot be assumed that all or part of an Inferred Mineral Resource will be upgraded to an Indicated or Measured Mineral Resource as a result of continued exploration.

Risk Adjustment

Precious metals deposits have skewed grade distributions. Skewed grade distributions have the property that a small proportion of samples can represent a disproportionately large amount of metal. The limited number of these samples can introduce significant uncertainty into a resource estimate. It is a common practice to cut the grades of very high-grade samples, restrict their projection distance or to adjust resource models to mitigate down-side risk.

In many precious metals deposits, Kupol included, the highest grade samples are scattered and discontinuous at the exploration drill-hole spacing. The number of high-grade samples intersected can vary according to the positioning of the drill holes, and it is impossible to know in advance which positions would give the most accurate estimate of the amount of high-grade metal actually present. The uncertainty related to the amount of high-grade metal can be evaluated using a Monte Carlo simulation technique developed by Mineral Resources Development/AMEC that has been applied over a 13-year period. This method essentially redrills the deposit 1000 times and notes the variation in the amount of high-grade metal present in annual or global production increments. The 20^th percentile of the simulated metal contents is added to the metal content represented by the remaining samples to give a risk-adjusted metal content. The difference between total metal content and risk-adjusted metal content is termed metal at risk. Theoretically, in four periods out of five, the mine should do better than the estimate; however there is additional and largely unquantifiable uncertainty related to the representivity of the sample-grade frequency distribution input to the simulation.

4

The appropriate time-period for Indicated Resources is annual, as Indicated Resources will be used to prepare annual production schedules as part of scoping and feasibility studies. For Inferred Resources, there is inadequate information to support annual planning; a global time-period is used.

The method has advantages over other top-cutting methods in that it takes into account 1) the data density, and 2) the volumes of increments used for production scheduling. As the data density is increased, the amount of metal at risk declines; longer production increments will have less risk than shorter ones.

At Kupol, frequency distributions of 1.5 m composite grades were inspected and classified into three groups (higher-grade Indicated, located in the Big Bend Sector; lower-grade Indicated, and Inferred). The composites were declustered and Monte Carlo Simulations run. The table below shows the results of these simulations for gold:

Area	Production Increment	No. Currently Available Composites Mined in period	High-grade Threshold (g/t Au)	Expected High- grade Composites in Period		Metal Represented	Metal at Risk
				No.	%
High-grade Indicated	Annual	113	110	7	5.8	28.5%	11.2%
Low-grade Indicated	Annual	113	50	5	4.8	27.1%	13.0%
Inferred	Global	571	60	37	6.5	28.8%	3.6%

The metal at risk is typical of scoping/prefeasibility stage projects. Typically with infill drilling (feasibility stage), the metal at risk for Indicated Resources is approximately 5 percent.

Metal at risk was removed by topcutting the grades of a resource model constructed from 1.5 m composites based on assays capped at 300 g/t Au and 2400 g/t Ag. The summary resource has been declared on a risk-adjusted basis. The risk-adjustment also tends to ameliorate the overprojection of high-grade intercepts. The tables below show a comparison of tonnages and grades based on a model that is totally uncut and the risk-adjusted model.

The Toronto Stock Exchange neither approves nor disapproves the information contained in this News Release. Bema Gold Corporation trades on The Toronto and American stock exchanges. Symbol: BGO.

Some of the statements contained in this release are forward-looking statements, such as estimates and statements that describe the Company’s future plans, objectives or goals, including words to the effect that the Company or management expects a stated condition or result to occur. Since forward-looking statements address future events and conditions, by their very nature, they involve inherent risks and uncertainties. Actual results in each case could differ materially from those currently anticipated in such statements.

5