Subjects for this study were 43 men convicted of spouse abuse and 40 men convicted of nonviolent crimes. All subjects were inmates in the Sacramento County (California) correctional system and were serving sentences of 1 year or less in the county jail. There were no significant differences between the groups on age, education, or ethnicity. Subjects were administered the Personality Assessment Inventory, Straus and Gelles' Conflict Tactics Scales, Griffin and Bartholomew's Relationship Styles Questionnaire, and Dutton's Borderline Personality Organization Scale. A variable relating to problems with anger control and whether the subject had an anger problem were taken from the structured clinical interview that was administered to each subject at the beginning of the data collection process. Barrett's Impulsiveness Scale was used to determine the impulsiveness of the groups. Impulsive Aggression was measured in the clinical interview by answers to six questions. The study found that impulsiveness and impulsive aggression correlated with measures of Borderline Personality Disorder and Antisocial Personality Disorder. In addition, the measure of Borderline and Antisocial Personality Disorder correlated significantly with physical aggression. The violent and nonviolent groups differed on impulsive aggression and on Borderline Personality Disorder. A partial replication of Tweed and Dutton's findings (1998) revealed sub-groups of high-violence and low-violence men. The high-violence group differed significantly from the low-violent and nonviolent groups. The high-violence group had higher pathology scores on all clinical scales of the Personality Assessment Inventory, except for Mania. The authors discuss the implications of these findings for violence prediction and the treatment of violent men. 5 tables and 31 references