The purpose of this project is to propose and test a model of anger escalation and de-escalation in couples with and without husband-to-wife aggression. This model integrates social learning, feminist, and systems theories. The following processes are hypothesized to maintain aggression. Relationships occur in the context of both blatant and subtle sexism. Even if the majority of aggressive men do not approve of aggression, they may still be sensitive to themes of disrespect by their wives. Due to skill deficits, aggressive couples will have a harder time navigating gender-based goal incongruence adaptively, leading to increased frequency of reciprocated hostility. As these negative reciprocity cycles occur more frequently, and are negatively reinforced, they become highly predictable and scripted. Appraisals and behaviors become overlearned, so that it only requires a subtle cue to set off a flood of anger. De-escalation deficiencies then make it difficult to stop the rush of anger once it starts. Spouses beliefs that they can resolve conflict become eroded, leading them to avoid these hot issues. Inevitably, these conflicts reoccur, leaving the couple even less prepared to effectively deal with such issues. With each repetition, the behaviors become more overlearned, coercion is more strongly reinforced, and global perceptions of the partner becomes increasingly negative. This process sets up the couple for husband-to- wife aggression, but typically produced only verbal aggression with the potential for further escalation. This model will be tested by comparing the escalation and de-escalation processes and skills of three groups of couples (n=50 per group): non- distressed/non-aggressive, distressed/non-aggressive, and distressed/aggressive. Participants will be recruited through random digit telephone surveying to increase the generalizability and ethnic diversity of the sample. Among the major hypotheses to be tested are (1) aggressive men demonstrate higher degrees of overlearning in their anger responses, as evidenced by (a) greater magnitude of anger escalation; (b) greater speed of anger escalation; and (c) predictable appraisal patterns; and (2) aggressive men have de-escalation skill deficits, as evidenced by (a) higher peak levels of anger; (b) extended periods of anger; (c) fewer attempts at de-escalation; and (d) fewer successful attempts at de-escalation. Improved treatment efficacy may be obtained by breaking the overlearned anger habit before attempting more cognitive interventions.