The study found that men who had perpetrated intimate partner violence (IPV) did not make more errors in assessing emotional states from the facial affect slides than did nonviolent (NV) men who had not engaged in intimate partner violence; however, the subtype of IPV perpetrator assessed as generally violent or antisocial (GVA) was less adept at accurately labeling slides of facial affect. On the other hand, the subtype of IPV abuser diagnosed as "borderline" (engaged in moderate to severe violence and showed the highest levels of emotional volatility, dependency, and psychological distress) were particularly accurate in identifying emotions from facial affect. The clinical implication of these findings is that training in emotional recognition skills may be useful for GVA batterers only. Study participants were 110 couples who responded to local newspaper adds seeking couples who were married or living together as if married for at least 6 months and were at least 18 years old. In order to meet preliminary screening for the domestically violent group, female partners had to report at least one incident of male-to-female aggression in the past year. To meet screening criteria for the NV group, women had to report no male-to-female violence in the past 5 years and no serious violence ever in their relationship. Only the males completed self-report measures of psychopathology. Other instruments administered to the men measured self-deceptive enhancement and impression management, borderline personality features, and history of violence. The men were asked to label 60 slides of facial expressions of happiness, surprise, fear, sadness, disgust, and anger, validated as recognizable across cultures. 2 tables and 47 references