Frequency was measured by reported frequencies of aggressive behaviors during the preceding year, by whether respondents could recall serious disputes, and by the recency of the disputes they recalled. In support of studies that have relied on official arrest statistics, the evidence suggests that ex-offenders engage in violence with greater frequency than the other two groups and that they have a greater tendency to physically attack and injure their antagonists when involved in violent disputes. Ex-mental patients appear to use weapons more frequently than the general population, but they are no more likely to injure antagonists. In contrast to labeling theory, police were no more likely to become involved or to make an arrest in incidents involving ex-patients or ex-offenders. (Publisher abstract)