Data were obtained from three groups of respondents employed by a large, metropolitan police force. The role perceptions of 47 new patrolmen were measured four times over a period which extended from the job application (T1) to about 2 years on the job (T4). The role perceptions of 52 veteran patrolmen and the role expectations of 26 supervisors were also measured at T4. Role perceptions and role expectations were measured on a scale of perceived job involvement. Role ambiguity scores were derived by subtracting the role perceptions of the patrolmen from the normative role expectations of the supervisors. Principal component analyses with varimax rotations of two components revealed that the factor structures of the measures of role perception and role ambiguity were constant over time. Analyses of the longitudinal trends in the data showed that a patrolman's role perceptions and his role ambiguity significantly decreased over the first 2 years of work. Moreover, stepwise discriminant function analyses showed that role perceptions distinguished high performers from low performers. High performers had lower role perceptions scores (lower perceived job involvement) than did low performers. This suggests that a norm of inactivity existed in the department examined. Because of this norm, the patrolmen who were rated as the best performers were those who spent less time and energy on the job. A major concern of urban citizens and police administrators should be to replace this inactivity norm with one which is more constructive. Instruments and other materials used in the study are appended. Tabular data and approximately 40 references are provided. (Author abstract modified)