May 7, 2012

Re:                               Theravance’s 2012 Annual Meeting of Stockholders — May 15, 2012

Proposal 3, Advisory Vote Regarding Executive Compensation (“Say on Pay”)

Dear Stockholder:

By now you should have received Theravance’s Notice of the 2012 Annual Meeting and Proxy Statement.  You can also view our Proxy Statement at http://investor.theravance.com/secfiling.cfm?filingID=1047469-12-4297&CIK=1080014.

We are writing to ask for your support at the Annual Meeting by voting in accordance with the recommendations of our Board of Directors on all proposals.  In particular, we want to request your support on Proposal 3, Advisory Vote Regarding Executive Compensation (“Say on Pay”).

Institutional Shareholder Services (ISS) has recommended that Theravance stockholders vote against this proposal.  We disagree with the ISS recommendation and believe that it reflects a general lack of understanding about drug research and development companies and a specific lack of understanding about both our business model and our CEO compensation program.

Our position can be summarized as follows:

·                  ISS based its conclusion that Theravance’s CEO pay was above market using a peer group selected based on revenue size.  We believe market capitalization is a much more meaningful measure of company size for our industry because it reflects stockholders’ projected value of the product pipeline of research and development-stage drug companies. Our business model emphasizes future relative profitability over current revenue.  By employing partnership agreements where the primary economic benefit accruing to our company is royalty income, our current and near-term future revenues will be lower than if we utilized a business model that sought to maximize absolute revenues. Our investors value Theravance based on future royalty streams as opposed to absolute revenue.  Given our business model, market capitalization is a much better metric than revenue for peer group selection.

·                  ISS inflated Theravance’s 2011 CEO pay using a flawed methodology, which: (1) attributed a larger-than-normal initial equity award that is the first part of a three-year equity program to a single year, and (2) valued a six-year performance-contingent equity award at its maximum potential value, while we consider the possibility remote that sufficient performance goals will be achieved for the award to fully vest.  In accordance with SEC rules, no value related to this performance-contingent award was included in the 2011 Summary Compensation Table in our proxy statement as it is not probable that the full amount of this award will be earned.

1

·                  Theravance would not have any “high” concern ratings under ISS’ CEO Pay for Performance Test, if either: (1) Theravance’s internal peer group were used to evaluate our pay and performance, or (2) Theravance’s CEO’s 2011 compensation were valued in a manner taking into account the high risk associated with the performance-contingent awards, and compared to ISS’ flawed peer group.

Why the ISS Peer Group is Not Appropriate

We take issue with the “one-size-fits all” peer group selected by ISS to benchmark our CEO compensation.  We believe that it fails to recognize that investors in drug research and development companies do not use revenue as a key measure of organizational size, scope, and complexity.  It further ignores the fact that Theravance has a market capitalization value that is over eight times larger than the median peer used by ISS in its analysis, and that this higher market value naturally drives higher equity grant value.

We agree that company size matters in setting compensation, and we consider the market value ascribed by our investors to the future value of Theravance’s product pipeline, for which management is responsible, as a better size measure.  Therefore, we developed our peer group (with advice from our independent compensation consultant) based primarily on market capitalization, development stage and labor market, and set CEO compensation using the pay practices of these companies as a guide.

This difference in comparator groups is the core source of the difference between the recommendations of our Board and ISS on the Say on Pay proposal, and as shown below (and in the Appendix with more detail), ISS chose to evaluate our CEO’s pay against a group of companies that we believe are smaller and generally less mature enterprises.

2

Market capitalization, a key component of which is stock price, is the primary driver of equity compensation grant value, and equity compensation grant value is the single largest component of CEO compensation at Theravance and the peer group. Obviously, higher market capitalization companies have higher equity compensation grant values because their stock price is higher, even if the percentage of stockholder value offered to management as equity compensation is not different.  For example, assume companies A and B have identical revenues of $25 million and both have 100 million shares outstanding, but company A’s market capitalization is $230 million (i.e., a $2.30 per share stock price) as compared to company B’s $1.6 billion market capitalization (i.e., a $16 per share stock price).  It is logical that company B would provide more annual equity grant value to its CEO because it has a higher price at the time of grant.  However, the ISS analysis, which effectively compares Theravance to peers with median market cap 1/8th our size, suggests that both companies should have the same grant value because both companies have the same revenues. This simply doesn’t make sense.

If ISS had used Theravance’s internal compensation peer group, then we would have passed the ISS CEO compensation tests without being in the “high” concern range on any quantitative measure.  This is true even using ISS’ flawed valuation of our CEO’s pay (discussed in more detail below), which failed to annualize the front-loaded restricted stock award amount over the three year period and valued a challenging six-year performance-contingent award at the maximum possible earn-out (this award was not expected to fund at the time of grant, and Theravance still considers the likelihood of achieving sufficient milestones to earn the full award as remote).

ISS’ Flawed Quantitative Pay Analysis

With respect to ISS’ quantitative pay analysis, we note the following methodological flaws:

·                  ISS uses mostly smaller market cap companies in its analysis (the median market cap is 1/8th our size), so it is no surprise that our CEO’s compensation was viewed as being relatively higher.

·                  ISS values a multi-year front-loaded grant of restricted stock awards (RSAs) as one year of compensation for our CEO, when in fact it is intended as roughly two years of compensation (in a three-year program), supported by an explicit written commitment not to grant our CEO more equity than contemplated under the original three-year program through 2012 and 2013.  As disclosed in our proxy statement, the intention of the Compensation Committee was to provide the equivalent of 110,000 RSAs in 2011, 2012, and 2013, resulting in a three-year total of 330,000 RSAs.  To augment retention, this was delivered as 220,000 RSAs in 2011 (with an additional fifth year of vesting).  Further, instead of granting an additional 55,000 RSAs in both 2012 and 2013, for a total of 330,000 RSAs over three

3

years, we reduced our CEO’s 2012 and 2013 awards by 33,000 shares each year, requiring 66,000 RSAs to be deducted as a “buy-in” for participation in a six-year performance plan.  The three- and six-year compensation designs are consistent with Theravance’s multi-year drug research and development business timeline, but if these awards were treated as a repeating annual event (which they are clearly disclosed as not so being), then CEO compensation is overstated significantly.  Theravance’s perspective on the long-term view of our CEO’s 2011 through 2013 compensation is that his compensation is equivalent to 110,000 RSAs per year, yet ISS treats 2011 as 220,000 RSAs (plus maximum potential performance-contingent award earn-out on top).

·                  ISS treats our CEO’s one-time six-year performance-contingent restricted stock grant, some or all of which may be forfeited if challenging performance goals are not achieved, as “annual” compensation in 2011.  Further, ISS has valued this performance-contingent grant as though the maximum possible share earn-out will be achieved.  In contrast, the probability of achieving any earn-out under the performance-contingent award was considered low enough that no expense was booked in our audited 2011 financials, nor in the first quarter of 2012, nor was any grant date compensation reported in the Summary Compensation Table of our proxy statement.

·                  ISS’ rating of “high concern” on the CEO pay as a multiple of median test generated its negative recommendation.  However, if this test had properly treated Theravance’s annual CEO grant as 110,000 RSAs, then Theravance’s CEO pay multiple would be 2.3x the median of ISS’ flawed group of too small companies, which would be in the “low concern” range under their published policy guidelines.

·                  A simulation of the ISS quantitative CEO pay for performance test, using Theravance’s internal compensation peer group, is shown in the Appendix.  The simulation indicates that even if the CEO’s entire 2011 front-loaded RSA and the maximum earn-out of the performance-contingent award are treated as “annual” compensation for 2011, then Theravance’s CEO pay is still below the “high concern” range.

ISS and Theravance stockholders were aware of the 2011 CEO front-loaded RSA and performance-contingent awards last year, as well as Theravance’s 75th percentile philosophy and internal peer group.  All of this information was disclosed in the March 16, 2011 proxy statement and none of it has changed since then.  Meanwhile, ISS supported the Theravance pay program in 2011, as did a substantial majority of the stockholders that voted on the 2011 proposal (98.6% of the vote was “for” the executive pay program).  The only changes to Theravance’s CEO equity grant program since the 2011 proxy statement disclosure were to make the program more stockholder-friendly by (1) committing that CEO equity grants would not exceed 22,000 RSAs in each of 2012 and 2013, and (2) making 50% of the CEO’s 2012 RSA award performance-contingent, which was not contemplated in the original three-year program design.

We would also like to reiterate, that 2012 and 2013 CEO compensation will appear to be quite a bit lower than 2011, since the front-loaded structure of the equity grant program causes the most recent proxy statement to show high compensation followed by considerably lower

4

amounts and we have committed not to exceed the original planned grant amounts for our CEO.

Theravance has Demonstrated its Commitment to Best Practice Performance-Based Executive Compensation and Governance Design.

·                  In fiscal 2011, 60% of executive officer equity awards were performance-based and we consider the performance goals applicable to those awards to be very challenging.

·                  Approximately 58% of our CEO’s equity awards granted since 2007 have been performance-based, and the goals established have been so challenging that 80% of those performance-based awards with a completed performance period were not earned and were forfeited as a result.  Theravance implemented pure performance-based equity awards several years ago and the Compensation Committee lived by the terms of those awards, including forfeiture of the shares when the goals were not met.

·                  We adopted an executive stock ownership guideline covering all named executive officers in early 2012, including a CEO ownership guideline set at six times annual base salary.

·                  We adopted a cash incentive compensation recoupment (“claw back”) policy effective February 8, 2012, and we intend to amend the policy to comply with the additional requirements of the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 (Dodd-Frank Act) after the SEC adopts new regulations for implementation.

·                  Our insider trading policy for executive officers and non-employee directors prohibits transactions involving short sales and derivatives, including put and call options and forward sales contracts.

The Appendix to this letter provides additional detail to support the conclusions herein.

We invite you to read the Proxy Statement for more information regarding the reasons the Board is recommending a vote “FOR” Proposal 3, Advisory Vote Regarding Executive Compensation.

We appreciate your time and consideration on these matters and ask for your support of the Board’s recommendation.

5

APPENDIX

If Theravance’s 2011 internal compensation peer group is used for ISS’ CEO relative alignment pay-for-performance test, Theravance will fall below ISS’ “high concern” range.  For the sake of conservatism, this analysis assumes that all CEO RSAs and the maximum earn-out under the performance-contingent award are treated as annual compensation for 2011 (as ISS did it in its report, although inconsistent with Theravance’s internal view).

2

Theravance is slightly below the median of its internal compensation peers when measured by market capitalization. However, compared to the comparator companies used by ISS in its evaluation of the CEO’s 2011 compensation, Theravance’s market capitalization is significantly larger.  Note that Theravance’s market capitalization size is approximately 8.5x larger than the median market cap in ISS’ compensation peer group.

7