Holding cash has a cost. For an EPS-maximizing CEO, that cost equals her firm’s earnings yield. EPS maximizers retain cash when they can get an even higher yield by investing the money. Otherwise, they return cash to shareholders. This is the EPS-maximizing payout policy. Growth stocks (EY < rf) never return cash because they can clear their low earnings-yield hurdle by investing in risk-free bonds. Value stocks (EY > rf) face a higher hurdle, which makes cash their cheapest source of capital but also raises the opportunity cost of retention. Value stocks return cash when they cannot invest in enough high-yield projects to make up for the higher cost. Paradoxically, the firms that are most likely to distribute cash—deep value stocks (EY ≫ rf)—are also the ones for whom internal cash is cheapest relative to external financing. Dividends and buybacks both deliver the same value to shareholders, but only buybacks can be accretive. Hence, EPS-maximizing CEOs prefer to distribute via buybacks. Some firms pay dividends for reasons outside our model. Such departures should concentrate among marginal value stocks (EY ≈ rf + ε) where the accretive pull of buybacks is weakest. Empirically, this logic explains which firms return cash, how they distribute the money, and time-series trends.<br><br>Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at <a href="http://www.nber.org/papers/&#119;34960" TARGET="_blank">www.nber.org</a>.<br>