The level of aggregate excess reserves held by U.S. depository institutions increased significantly at the peak of the 2007-09 financial crisis. Although the amount of aggregate reserves is deter- mined almost entirely by the policy initiatives of the central bank that act on the asset side of its balance sheet, the motivations of individual banks in accumulating reserves differ and respond to the impact of changes in the economic environment on individual institutions. We undertake a systematic analysis of this massive accumulation of excess reserves using bank-level data for more than 7,000 commercial banks and almost 1,000 savings institutions during the U.S. financial crisis. We propose a testable stochastic model of reserves determination when interest is paid on reserves, which we estimate using bank-level data and censored regression methods. We find evidence primarily of a precautionary motive for reserves accumulation with some notable heterogeneity in the response of reserves accumulation to external and internal factors of the largest banks com- pared with smaller banks. We combine propensity score matching and a difference-in-differences approach to determine whether the beneficiaries of the Capital Purchase Program of the Troubled Asset Relief Program accumulated less cash, including reserves, than non-beneficiaries. Contrary to anecdotal evidence, we find that banks that participated in the program accumulated less cash, including reserves, than nonparticipants in the initial quarters after the capital injection.