Page 1

Directors:

A J Wright (Chairman), I D Cockerill

†

(Chief Executive Officer), K Ansah

, A Grigorian°, N J Holland

†

(Chief Financial Officer), J G

Hopwood,
G Marcus, J M McMahon

†

, D M J Ncube, R L Pennant-Rea

†

, P J Ryan, T M G Sexwale, C I von Christierson

†

British,

Ghanaian, °Russian

Corporate Secretary:

C Farrel

Gold Fields Limited

Reg. 1968/004880/06
24 St Andrews Road
Parktown, 2193

Postnet Suite 252
Private Bag X30500
Houghton, 2041
South Africa

Tel +27 11 644-2400
Dir +27 11 644-2502
Fax +27 11 484-0590

www.goldfields.co.za

EXECUTIVE

Karl Hiller
Branch Chief
United States Securities and Exchange Commission
Division of Corporation Finance
100 F Street, N.E.
Washington, D.C. 20549-7010
United States of America

8 June 2007

By EDGAR

Dear Mr. Hiller:

Re: Gold Fields Limited

Form 20-F for the year ended June 30, 2006

Filed on November 24, 2006

File No. 001-31318

We refer to the comment letter (the “Comment Letter”) dated May 17, 2007 of the staff (“Staff”) of the
Division of Corporation Finance of the Securities and Exchange Commission (the “Commission”) on
the above referenced Form 20-F (the “2006 Form 20-F”) of Gold Fields Limited (“Gold Fields” or the
“Company”). For your convenience, each comment is repeated below, prior to the response.
Management has discussed the content of this letter (the “Response Letter”) with its auditors.

Form 20-F for the Fiscal Year Ended June 30, 2006

General

Please amend your current Form 20-F to incorporate all of the revisions you have proposed to
make in response to prior comments 1, 3, 4 and 5.

Response

The Company acknowledges the Staff’s comment and, as discussed with the Staff in the
conference call on May 29, 2007, the Company intends to amend the 2006 Form 20-F once all
the Staff’s comments necessitating revisions to its current Form 20-F have been resolved.

Results of Operations – Years Ended June 30, 2006 and 2005, page 145

Depreciation and Amortization, page 150

We have read your response to prior comment 2, explaining that you changed your method of
calculating amortization to apply an amortization rate that will based on reserve estimates
prepared as of the end of the second quarter of the preceding fiscal year (December 31) to
production for the first nine months of the current fiscal year, and an amortization rate that will
be based on reserve estimates as of the end of the second quarter of the current fiscal year to
production for the last quarter of the current fiscal year. We understand that you previously

utilized a single amortization rate of the entire fiscal year, based on your reserve estimates as
of the end of the preceding fiscal year (June 30). You describe this change as having been
implemented in connection with your decision to no longer report mineral reserve information
that is current as of the end of the fiscal year covered by your annual report on Form 20-F.

Amortization Policy

While we understand that you may have established conventions which facilitate the process of
calculating periodic units-of-production amortization, under U.S. GAAP you need to update the
reserve estimates utilized in your formulas to reflect material information arising during the
period, of which you become aware, including significant changes in your reserves, such as
those resulting from acquisitions, dispositions or discoveries; and material changes in the
assumptions underlying your reserve estimates, such as changes in your mine plan, actual
yields that are inconsistent with prior assumptions, or changes in commodity prices or
production costs, as indicated in the guidance on estimates in APB 20 and SFAS 154. You
would also need to update your reserve information as of your current fiscal year-end to
appropriately conduct impairment testing, following the guidance in paragraphs 16 through 21
of SFAS 144.

Please implement procedures for monitoring changes in these and other factors underlying
your reserve estimates; assessing the implications of such changes on your reserve estimates,
amortization computations and views on recoverability; and adjusting your amortization
formulas and property account balances as necessary to take into account all changes having
material effects in the period that such changes occur and in which you become aware of the
information. Please submit an analysis showing haw the amortization expense you recorded
each period would need to change to properly reflect material revisions in reserves and the
related assumptions based on new information arising during the period, in advance of your
scheduled reserve updates. Similarly, advise us of your views on asset recoverability, after
taking into consideration all information that was known about your reserves on the balance
sheet dates. Additionally, expand your accounting policy disclosure to discuss the procedures
you implement.

Mineral Reserves

With regard to your preference for reporting non-current information about reserves in the Form
20-F, we ask that you read the guidance in General Instruction C(b), stating that “Unless an
item directs you to provide information as of a specific date or for a specific period, …give the
information in an annual report as of the latest practicable date.” In addition, you may find the
following guidance within Industry Guide 7, applicable by way of the Instruction to Item 4 of
Form 20-F, helpful in understanding the reporting requirement for reserve information:

Paragraph (b)(4)(i) of Industry Guide 7 requires a “…description of the present condition
of the property, the work completed by the registrant on the property, the registrant’s
proposed program of exploration and development, and the current state of exploration
and/or development of the property.”

Paragraph (b)(5) of Industry Guide 7 requires a description of the “…existing or
potential economic significance on the property, including the identity of the principal
metallic or other constituents insofar as known,” also stating that if reserves have been

established, disclose “…the estimated tonnages and grades (or quality, where
appropriate) of such classes of reserves....”

Please also read Instruction 1(a) to Item 4.D of Form 20-F, requiring disclosure of “…material
information about production, reserves, locations, developments and the nature of interest. If
individual properties are of major significance to you, provide more detailed information about
those properties....”

We do not believe that General Instruction C(b), or the Industry Guide 7 references to “present”
condition and “existing” economic significance accommodate a practice of not disclosing
information that is current as of the end of the period covered by your report; and we do not see
support for a view that reserves as of the end of the period, and changes in reserves during the
period, would not constitute material information requiring disclosure pursuant to Item 4.D.
Given the interrelatedness of accounting and mineral reserve measurements necessary to
comply with U.S. GAAP, coupled with the specific disclosure requirements cited above, we
again ask that you amend your filing to update your reserve information through June 30, 2006,
the end of your fiscal year.

Response

The Company acknowledges the Staff’s comments and wishes to advise the Staff that, as
discussed with the Staff in the conference call on May 29, 2007, it is formulating an approach to
address these issues. The Company expects to submit a separate letter outlining that approach
and otherwise responding to these comments shortly.

Engineering Comments

Information on the Company, page 32

Reserves of Gold Fields as of December 31, 2005, page 39

We have reviewed the reserve reports and feasibility study you supplied as supplemental
information in response to prior comment 6. Based on this review and your reserve redefinition
experience in the last couple of years, we believe that further disclosure is needed concerning
you reserve estimation approaches, including the basis you have for assuming continuity
between observation points in establishing probable reserves. Please expand your disclosure
of your estimation procedures for the proven, probable above infrastructure, and probable
below infrastructure categories of your underground South Africa reserves to address the
following points:

•

Detail your reserve estimation methodologies for each reserve category, highlighting
any significant differences in these methodologies from underground mine to mine, and
between reserve categories.

•

Compare and contrast the methodologies as to their reliability, citing your reserve
reconciliation experience between various categories of reserve estimates and actual
material delivered to the mill.

•

Detail primary sources of error in your assumption of continuity for your probable
reserves, both above and below infrastructure, with emphasis on the sources of error
which have led to significant reductions in reserves.

•

In a table, list the average spacing between drill pierce points or sample points for each
category of reserves, and the maximum distance that data is projected for each mine.

•

Explain how you have changed you practice of assuming continuity between the
surrounding observations in response to mining results precipitating and downward
revisions in you reserve estimates.

Response

The Company acknowledges the Staff’s comments and proposes to revise its disclosure by
inserting the following language on page 42 of the 2006 20-F, immediately after its gold ore
reserve statement.

“Additional information regarding reserves at the South African underground operations

The following sets forth the reserve estimation methodologies for the different categories of
reserves at the underground operations of each of the South African mines.

Driefontein

Reserve

Classification

Sample Spacing Range Min/Max

(meters)

Maximum Distance Data is Projected

(meters)

Proven

3 to 180

110

Probable

(AI)

(1)

3 to 1,140

570

Probable

(BI)

(1)

3 to 2,840

1,420

Note:
(1)

AI is above infrastructure; BI is below infrastructure.

For proven reserves, the orebody is opened-up and sampled on a 3 meter spacing for
development (such as raises), and a 5 meter grid for stoping, together with underground
borehole spacings ranging from tens to hundred of meters. Blocks classified as proven are
therefore generally adjacent to close spaced sampling and generally pierced by a relatively
dense irregular pattern of boreholes. Estimation is constrained within both geologically
homogenous structural and facies zones, and is generally derived from either ordinary or simple
kriged small-scale grids, ranging from 10 meter to 20 meter block sizes.

For above infrastructure probable reserves, the estimates access the vast numbers of opened-
up samples on a 3 meter spacing for development, and a 5 meter grid for stoping bordering
these areas. In addition underground borehole spacings ranging from tens to hundred of meters
are used together with surface drillholes and seismic surveys. Blocks classified as probable
(above-infrastructure) are generally adjacent to blocks classified as proven. Estimation is
constrained within both homogenous structural and facies zones, and is generally derived from
either ordinary or simple kriged medium to macro scale sized grids ranging from 40 meter to
420 meter sizes, or through declusted averaging or Sichel ‘t’ techniques. For planning purposes
these blocks are further evaluated to facilitate the selection of blocks above the cut-off grade.

For below infrastructure probable reserves, the estimates access the vast numbers of opened-
up samples on a 3 meter spacing for development, and a 5 meter grid for stoping bordering
these areas. In addition underground borehole spacings ranging from tens to hundred of meters

are used together with surface drillholes and seismic surveys. Blocks classified as probable
(below infrastructure) are therefore generally below blocks classified as proven or probable
above infrastructure. Estimation is constrained within both homogenous structural and facies
zones, and is generally derived from either ordinary or simple kriged medium to macro scale
sized grids ranging from 40 meters to 420 meter sizes, or through declusted averaging or Sichel
‘t’ techniques. For planning purposes these blocks are further evaluated to facilitate the
selection of blocks above the cut-off grade.

Generally, there are no major differences in estimation techniques between probable reserves
classified as above infrastructure and below infrastructure, although above infrastructure
probable reserves are adjacent to close spaced mining data while below infrastructure probable
reserves are below close spaced mining data. When viewed in isolation, the below
infrastructure probable reserves may include different borehole spacings, maximum projection
distances and estimation block sizes.

Kloof

Reserve Classification

Sample Spacing Range Min/Max

(meters)

Maximum Distance Data is Projected

(meters)

Proven

3 to 150

150

Probable (AI)

(1)

3 to 718

360

Probable (BI)

(1)

3 to 1,390

890

Note:
(1)

AI is above infrastructure; BI is below infrastructure.

Estimations for proven reserves are made on the same basis as at Driefontein.

Estimations for above infrastructure probable reserves are made on the same basis as at Driefontein,
but with medium sized kriged grids starting from 40 meters to macro blocks of 400 meters. For
planning purposes these blocks are evaluated to facilitate the selection of blocks above the cut-off
grade.

Estimations for below infrastructure probable reserves are made on the same basis as at
Driefontein, but with medium sized kriged grids starting from 40 meters to macro blocks of 400
meters. The distinction between estimation techniques for above infrastructure and below
infrastructure probable reserves is the same as at Driefontein. For planning purposes these
blocks are evaluated to facilitate the selection of blocks above the cut-off grade.

Beatrix

Reserve Classification

Sample Spacing Range Min/Max

(meters)

Maximum Distance Data is Projected

(meters)

Proven

3 to 120

120

Probable (AI)

(1)

3 to 940

750

Probable (BI)

(1)

540 to 610

740

Note:
(1)

AI is above infrastructure; BI is below infrastructure.

Estimations for proven reserves are made on the same basis as at Driefontein but with kriging
blocks ranging from 16 meters to 32 meters.

Estimations for above infrastructure probable reserves are made on the same basis as at
Driefontein but with medium-sized kriged blocks of 32 meters, and macro geological zone
estimates being made through declusted averaging or Sichel ‘t’ techniques. For planning
purposes these blocks are evaluated to facilitate the selection of blocks above the cut-off grade.

Estimations for below infrastructure probable reserves are made on the same basis as at
Driefontein but with medium-sized kriged blocks being 32 meters, to macro geological zone
estimates through declusted averaging or Sichel ‘t’ techniques. The distinction between
estimation techniques for above infrastructure and below infrastructure probable reserves is the
same as at Driefontein. For planning purposes these blocks are further evaluated to facilitate
the selection of blocks above the cut-off grade.

Grade control and reconciliation practices at the South African operations follow similar
procedures to those applied elsewhere in Witwatersrand deep level gold operations. The reefs
and hangingwall and footwall lithologies are visually identifiable and channel sampling ensures
that the face grade is monitored accordingly. An exercise is undertaken to reconcile reserve
estimates with actual mined tonnages and grades. As part of the reconciliation exercises,
physical factors including channel width, stope width, dilution, mine call factor and block factor
are monitored and recorded on a monthly basis.

An annual measure of the estimation efficiency is made by means of the block factor. The block
factor is the ratio, expressed as a percentage, which the total specific mineral content of the ore
broken from blocks (as indicated by the follow-up results based on close spaced 3 meter and 5
meter development and mining sampling inside these blocks) bears to the content of this ore as
reflected in the corresponding planned kriged estimates for ore reserves.

The following table sets forth the actual versus estimated block factors at the South African
operations for the last three reserve declarations:

Block Factor (Actual)

Block Factor (Estimated)

June 30,

2004

June 30,

2005

December

31,

2005

June 30,

2004

June 30,

2005

December

31,

2005

Driefontein

98 100

Driefontein

100 99 99

Kloof

106 100

Kloof

94 100 99

Beatrix

101 101

101

Beatrix

100 100 100

The mine call factor is the ratio, expressed as a percentage, which the amount of gold
accounted for in the mill (recovery plus residue) bears to the corresponding amount of gold
called for by the mine’s measuring and evaluation methods. Sources of deviation are, among
other things, gold losses incurred in blasting, cleaning (preparing broken ore for transportation),
transportation and inefficient mining.

The following table sets for the actual versus estimated mine call factors at the South African
operations for the last three reserve declarations:

Mine Call Factor (Actual)

Mine Call Factor (Estimated)

June

30,

2004

June 30,

2005

December

31,

2005

June 30,

2004

June 30,

2005

December

31,

2005

Driefontein

89 89 92

Driefontein

90 90 89

Kloof

85 83 84

Kloof

87 86 87

Beatrix

81 82 97

Beatrix

89 86 87

The primary assumptions of continuity of the geologically homogenous zones are driven by the
geological model, which is updated only if new information arises. Any changes to the model
are subject to peer, internal technical corporate consultant and independent consultant review.
Historically, mining at South African deep level gold mines has shown significant geological
continuity, so that new mines were started based on limited surface borehole information.
Customarily, geological facies are primarily based on the definition of different facies within each
conglomerate horizon. These facies are extrapolated into new, undeveloped areas taking into
account any surface borehole data in those areas. Normally these facies are continuous,
supported by extensive historical sample databases, and can be incorporated in the macro
kriging of large blocks. However, deviations regarding geological facies continuity patterns
began to be observed in the down-dip area at Kloof Eastern Boundary Area, or EBA, prior to the
year ended June 30, 2005. Additional information accumulated over time from mining, seismic
surveys and drilling resulted in the modification of the prevailing geological model, particularly
below infrastructure, for the gold ore reserve statement as of June 30, 2005.

An external geological consultant review undertaken in 2005 endorsed the revised model and
highlighted areas for further refinement. The model used before the year ended June 30, 2005
had geological facies trends honoring the old geozones in the below infrastructure areas, that
is, along a west to east palaeocurrent (sediment input) direction. The external review confirmed
that in the below infrastructure area, younger sandy facies were introduced from a northern
source, with sediment transported along a north to south palaeocurrent direction, thus resulting
in a change in the geological interpretation for the Ventersdorp Contact Reef, or VCR, orebody,
particularly impacting the below infrastructure areas. The current EBA model encompasses the
findings of the review, with geozones for the sandy facies honoring this north-south
palaeocurrent direction and as a result, the maximum extrapolation distance of estimates from
data has been reduced.

These differences affected not only the entirety of the below infrastructure project at the EBA
but also, to a lesser degree, affected the below infrastructure project at the Kloof Extension
Area, or KEA, therefore also resulting in the revision of the KEA footprint, and thus reserves,
downwards.”

In addition, the Company wishes to advise the Staff that it has supplied certain supplemental
information regarding these issues to Ken Schuler, Mining Engineer, under separate cover.

As supplemental information, please provide example mine maps for each of your South
African mines that illustrate the area of influence of your sample points and the category
assigned to those areas. Also provide a copy of any corporate reserve estimation manuals or
guidelines.

Response

The Company acknowledges the Staff’s comment and wishes to advise the Staff that it will
supply the requested supplemental information to Ken Schuler, Mining Engineer, under
separate cover. If Mr. Schuler or other members of the Staff have any questions regarding this
information or the supplemental information provided in response to Comment 3, they should
feel free to contact Tim Rowland at +27 11 644 2476.

Supplemental information

Pursuant to a request from the Staff, Gold Fields hereby acknowledges (i) it is responsible for the
adequacy and accuracy of the disclosures in the filing, (ii) Staff comments or changes to disclosure in
response to Staff comments do not foreclose the Commission from taking any action with respect to
the filing and (iii) it may not assert Staff comments as a defense in any proceeding initiated by the
Commission or any person under the federal securities laws of the United States.

* * * * *

Should you or the Staff have any questions or require any additional information, please contact the
undersigned at +27 11 644 2502 or via e-mail at nholland@goldfields.co.za.

Yours sincerely,

/s/ Nicholas Holland

Nicholas J. Holland
Chief Financial Officer

Gold Fields Limited

cc:

Lily Dang, Securities and Exchange Commission

Jenifer Gallagher, Securities and Exchange Commission
Ken Schuler, Securities and Exchange Commission
Michael Fleischer, Gold Fields Limited
Paul Schmidt, Gold Fields Limited
Jennifer Schneck, Linklaters