The offender sample consisted of 32 male inmates (16 white, 16 black) of a minimum- to medium-security State correctional institution; the nonoffender sample was composed of 29 men (14 white, 15 black) who were similar to the offenders in age, IQ, and socioeconomic background, including a history of irregular employment. Offenders and nonoffenders were required to respond to eight interpersonal conflict situations, two situations designed to elicit positive assertive behaviors, and one situation that began as a positive situation and changed to one of conflict. Each participant also completed three self-report measures: The Adult Self-Expression Scale, the Buss-Durkee Hostility Inventory, and the Social Avoidance and Distress Scale. Multivariate analysis of variance revealed significantly higher aggression among offenders on role-playing measures. Although there was no significant univariate difference in assertiveness ratings, this result was complicated by significant and varying relationships between assertion and aggression in the two groups. With this relationship controlled, there was a higher level of rated assertion among nonoffenders. The findings suggest the advisability of including an emphasis on aggression reduction in assertiveness training programs for offenders and the importance of including measures of aggression in outcome studies of assertiveness training. Six tables and 43 references are provided. (Author abstract modified)