This study used logistic regression analysis to identify psychosocial predictors of unprotected anal intercourse in a sample of 212 HIV positive gay men who were enrolled in a sexual risk reduction intervention. A combination of five psychosocial variables (i.e., age, avoidant coping, loneliness, depressive symptoms, and impulsivity) reliably discriminated between men who had engaged in unprotected anal intercourse (UAI) with HIV negative or serostatus unknown partners and those who had not engaged in this high risk behavior. Avoidant coping emerged as the best psychosocial predictor that distinguished between the two groups. Men who had UAI during the previous 4 months had significantly higher scores on avoidant coping as compared with the no-UAI group. These findings suggest the need to design intervention programs for HIV positive gay men that addresses coping strategies in the context of high-risk sexual behaviors.