Hedge fund managers receive as performance fees a large fraction of their funds’ profits, in addition to regular fees proportional to funds’ assets. Performance fees are paid only when a fund exceeds its previous maximum - the high water mark. The most common scheme, dubbed Two and Twenty, entails performance fees of 202 We study the risk shifting incentives created by such fees, solving the portfolio choice problem of a manager with constant relative risk aversion, constant investment opportunities, and a long horizon. The portfolio that maximizes expected utility from future fees is constant, and coincides with a Merton portfolio with effective risk aversion equal to the weighted average of the manager’s true risk aversion and the myopic value of one, with the performance fee as the myopic weight. Moreover, the optimal portfolio coincides with that of an investor facing the constraint of a maximum drawdown less than one minus the performance fee, as a fraction of the last recorded maximum. Since performance fees modify a manager’s risk aversion, we investigate their potential as agency tools, solving a Stackelberg equilibrium between an investor and a manager. We find that an equilibrium exists only if both the manager and the investor have very low risk aversion. In all other cases, no equilibrium is consistent with positive performance fees.