The analysis used a behavioral model of police spending, adapted from a model developed by Bahl, Johnson, and Wasylenko. The model included equations for both compensation of police and police employment. A test of the model used 1979 data for 66 Standard Metropolitan Statistical Areas (SMSA) with central city populations of more than 100,000. The disparities in expenditures between cities and suburbs are due partly to the greater needs of cities and partly to institutional arrangements and policy choices. The unionization of city employees increases city police salaries and also affects suburban salaries. Higher crime rates in central cities explain part of their higher level of police employment, but do not have a significant effect in suburbs. Crime and population size exert important pressures on police spending in cities but not in suburbs. Suburban expenditures tend to be lower where there is more governmental fragmentation. The data do not support the view that governmental consolidation will lead to lower per capita expenditures. Data tables and 28 reference notes.