Two experimental conditions, a manualized cognitive-behavioral anger-control treatment incorporating empowerment strategies and a relapse-prevention treatment without the anger-control component, were compared to assess their impact on levels of trait anger and attributional styles of women recovering from alcohol and drug addiction. Participants were predominantly African American, low-socioeconomic-level women living in a residential facility in a major Midwestern city. Although significant changes were found for the anger-control group participants, sample size (anger control n = 8, relapse prevention n = 5) and participant characteristics call for caution when interpreting the results and limit the generalizability of the intervention. As a preliminary study, the findings are encouraging. Recommendations for future research and implications for social work practice are discussed. (Contains 5 tables.)