Objective: This study investigated the association between hostility and health and whether it is moderated by the quality of an individual's primary romantic relationship. Method: Longitudinal data were provided by 184 African Americans, including 166 women. Participants averaged 38 years old and were married or in long-term marriagelike relationships. Hostility and relationship quality were measured at the first assessment. Hostility was based on participants' responses to items tapping cynical attitudes about relationships. Relationship quality was based on trained observer ratings of videotaped couple interactions on behavioral scales reflecting warmth, support, and communication skills. At 2 assessments approximately 5 and 7 years later, participants provided health data. Health index scores were formed from responses to five scales of the SF-12 (Ware, Kosinski, & Keller, 1998) as well as to responses to questions about the number of chronic health conditions and the number of prescribed medications. Results: Stepwise regression analyses controlling for demographic variables and the earlier health score tested the main and interactive effects of hostility and relationship quality on longitudinal changes in health. Whereas no main effects were supported, the interaction of hostility and relationship quality was significant (p < .05). The form of the interaction was such that high-hostile individuals had better health outcomes if they were in a high-quality relationship. Conclusion: Hostile persons in high-quality relationships may be at less risk for negative health outcomes because they do not regularly experience the physiologic reactivity and adverse psychosocial outcomes that they would otherwise experience as a result of recurring interpersonal conflict.