Seventy-two men participated in the study: 47 nonabusive subjects and 25 abusive subjects. The nonabusive participants were recruited through announcements in the community for participation in a marital study for men married a minimum of 2 years. The abusive participants included those in a court-mandated therapy program (n=6) and those who responded to announcements at clinics and reported an aggressive act against their partner in the previous 12 months (n=19) as assessed by the Conflict Tactics Scale. The study hypothesized that control is an important variable in understanding men's physical aggression against their partners. Subjects completed questionnaires that measured self-esteem, desirability of control, and perceived personal and interpersonal control. "Perceived control" refers to an individual's belief about his ability to alter events significantly independent of his actual ability to alter events. The "desirability of control" is the general desire for control over events in one's life. Logistic regression analyses showed a significant three-way interaction among desirability of control, perceived personal control, and self-esteem. Further examination revealed two subgroups of men at high risk for engaging in domestic violence: men low on self-esteem, low on desirability of control, and low on perceived personal control; and men high on self-esteem, high on desirability of control, but low on perceived personal control. Treatment implications are discussed for the type of abusers identified. 3 tables and 23 references