The three types of violent men studied were the generally violent/antisocial type, the dysphoric/borderline, and the family only/non-pathological partner abusers. The generally violent/antisocial type experienced the most severe childhood physical abuse and engaged in moderate to severe battering including psychological and sexual abuse. The dysphoric/borderline experienced the most severe parental rejection and moderate exposure to interparental violence. The family only/non-pathological partner abusers typically had experienced the least childhood trauma. The 153 (114 intimate violent and 39 non-intimate violent) participants were on probation for either intimate violence or non-intimate violence. In addition to being administered the MMPI-2, each of the participants was individually interviewed about the nature of their offense, criminal history, family history, work history, and a psychosocial history. The results confirmed the theoretical model proposed by Holtzworth-Munroe and Stuart (1994) regarding psychopathology as measured by the MMPI-2. The differences among the three groups on severity of violence were consistent with the tripartite model (lower severe violence for the non-pathological group than the other two groups). The tripartite model would suggest that the less the psychopathology, the less severe the violence. This was consistent with the results. The results failed to find statistically significant differences regarding family-only violence or exposure to family of origin violence, although these variables had frequencies/percentages similar to what would be expected for individual typologies of the tripartite model for some but not all three typologies. All three groups reported moderately high to high percentages of no exposure to family of origin violence. The clear differences between the men based on personality characteristics and the differences on severity of violence would appear to have direct implications for treatment. There were at least two or three separate groups of violent men in the sample. The importance of treatment matching would seem to be a logical conclusion from these results. 4 tables, 66 references