Data pertaining to a sergeant promotion process was collected from a sample of 30 officers, aged 29 to 46, from a medium-sized, urban police department. The two dependent variables included a peer evaluation of each subject by the other participants and a supervisor performance appraisal with promotion potential evaluation. Demographic and psychological characteristics, management style, communications style, use of power, and amount of job burnout variables were selected to represent a wide range of independent measures relevant to performance. The study found no consistent pattern of underlying constructs; the correlation between supervisor rankings and peer rankings was quite small. This analysis illustrates the definitional or validity problems of police performance evaluations. One way to solve these problems is to focus directly on task performance evaluation through job analysis or through critical incident techniques rather than on the underlying traits or dimensions that management believes are related to good performance. 2 tables and 32 references (Author abstract modified)