We compare two large-scale policy interventions aimed at reducing crime: subsidizing high school completion and increasing the length of prison sentences. To this purpose we use a life-cycle model with endogenous education and crime choices. We apply the model to property crime and calibrate it to U.S. data. We find that targeting crime reductions through increases in high school graduation rates entails large efficiency and welfare gains. These gains are absent if the same crime reduction is achieved by increasing the length of sentences. We also find that general equilibrium effects explain roughly one half of the reduction in crime from subsidizing high school.