Municipal police deployment involves a fundamental location tradeoff: more patrol in one place means less elsewhere. I study the distributional consequences in Chicago using granular data, quasi-experimental variation, and a quantitative urban model. The observed allocation is consistent with a dual objective of crime minimization and land value maximization, generating disproportionate patrol per crime in high-income neighborhoods. Shifting toward pure crime minimization reduces citywide crime but hurts every resident type. In neighborhoods gaining police, rising rents erode welfare gains for residents with the lowest willingness to pay for crime reduction, while in neighborhoods losing police, higher crime compounds losses for those with the highest willingness to pay. Residential segregation amplifies welfare disparities across types. A modest reallocation toward neighborhoods with high marginal valuation of crime reduction improves welfare for all types. Shifting the citywide police level generates limited welfare gains, pointing to reallocation as the main policy lever.