The effectiveness of community policing has yet to be determined, and social science and evaluation research are turning to this issue. Implicit in the approach of community policing is a belief that the values of social science should guide social decisionmaking, that this is a specialized task for trained outside evaluators, that crime is the most critical performance dimension, and that programs rather than organizations are the proper units of analysis. However, this framework may hinder the full development of community policing agencies as learning organizations. In addition, dimensions other than crime are important in evaluating police agencies, and evaluations of community policing require attention to organizations as well as programs. Research should focus on designing appropriate performance measures and monitoring systems and work with public agencies to accelerate the leading-edge police agencies into full-fledged community-policing departments. Evaluations should also focus on the administrative innovations and processes of institutional change that support and promote effective community policing and on interventions that are potentially both powerful and robust in a range of circumstances. Footnotes and 20 references (Author abstract modified)