Based on rational criminal behavior as exemplified in retail store burglaries, an empirical model is constructed which responds to the expected costs imposed on the criminal. Preventive patrol is designed to raise these expected costs by raising the probability of arrest. The analysis is based on the assumption that no change in preventive patrol strategy takes place during the data collection period. Changes in inputs can then be properly measured by changes in time devoted to preventive patrol. Employing statistical analysis techniques such as location analysis, linear discriminant function analysis, and aggregated regression tests, the model describes and evaluates the current activities of police patrol and detective forces as well as the probability of burglary activities in any given location. Two preventive patrol models are developed which concentrate on the use of search theory to design an optimal preventive patrol strategy. They combine an allocation model determining optimal coverage rates and a random patrol schedule model to evolve optimal patrol schedules which can be computed by mathematical programming. Two appendixes describe the mathematical models of police allocation to areas in the city to combat street crime and the preventive patrol decision process. Footnotes and 24 references are included.