The adoption of management science models has typically been studied in time frames of less than two years. Recommendations of government decisionmakers for improving the likelihood of successful model development, diffusion, and adoption have been based on these studies. Yet some factors important to the use of models do not emerge in the short term. This study examines model use over a longer time frame, and draws some policy implications. A survey was taken of 17 police departments live years after they implemented PCAM, the Patrol Car Allocation Model. Utilization (extent of use) and institutionalization (extent of organizational adaptation) were evaluated. Overall, there were few signs that use of PCAM had a major impact on the state of police patrol allocation. PCAM was found to be operational in six departments, and fully institutionalized in two. Based on patterns of use and nonuse, recommendations are made to enhance the long-term impact of future modeling projects. Among them are to survey user needs and give them a larger role in model development, to design models capable of an expanded range of tasks and of running on personal computers, and to include performance evaluation and feedback in all model-building projects. A structured approach to development is suggested to anticipate implementation and institutionalization problems, and to produce integrated solutions.