Domestic violence continues to pose a significant public health threat; some researchers contend that domestic violence acts are among the most frequently committed crimes. However, little is known about how relationship characteristics contribute to violence. Previous research has found evidence of problem-solving deficits among male perpetrators of spousal violence, but little is known about how the interaction style of the male contributes to violence in the home. The current study investigated how different types of situations interact with inter- and intrapersonal characteristics to produce aggressive responses in spouses. Participants were 12 maritally violent couples recruited from a violence reduction program for couples; the male in each dyad had been convicted of spousal assault and court-ordered to participate in the program. Participants completed a series of questionnaires, including the Anger Strategies Scale, the Couple Argument Severity Scale, the Social Problem Solving Inventory, and the Trait Anger Scale. Participants were then videotaped discussing three different vignettes involving neutral, low-conflict, and high-conflict situations. Three raters independently coded the videotapes, measuring changes in the communication skills of the couples using several categories from the Marital Interaction Coding System. Levels of anger, problem-solving skills, strategies for dealing with high-conflict situations, and gender differences were similarly measured. Results of statistical analyses failed to find significant differences between perpetrators and victims in this study in terms of social problem-solving skills, trait anger, and communication strategies. Effective and ineffective use of communication skills was influenced by the topic being discussed; ineffective skills dominated discussions of high-conflict situations. Many domestic violence programs focus on teaching coping and communication skills; the findings from this study suggest that programs should teach skills that are designed for managing high-conflict scenarios. Limitations of the study include problems of generalizability and a lack of a nonviolent control group. Tables, references