The author develops a theoretical model of bank closure. The regulatory decision about bank failure consists of two parts: whether to close and how to close. Assuming that the closure decision is credible, the welfare implications of two resolution regimes are considered. In one case, a meta-regulator supervises, closes, and resolves failed banks using an ex post efficient criterion. In the other case, a supervisor closes the bank while a deposit insurer resolves the closure on the basis of least cost. The bank chooses the riskiness of its loan portfolio in response to the announced policy. The supervisor can limit risk-shifting incentives of banks ex ante by raising capital requirements. The cost of this decision is a misallocation of economic resources, since some welfare-enhancing projects are abandoned. Market discipline is determined both exogenously, by the level of uninsured depositors, and endogeneously, by the regime and capital requirements chosen. Least costly resolution weakly dominates an ex post efficient resolution decision when market discipline is present. Neither mechanism outperforms when it is reliant on capital regulation in the absence of market discipline.