The purpose of this study was to determine the relative importance and interactive effects of partnership characteristics in unprotected intercourse among people living with HIV/AIDS (PLWHA). An interview study was conducted among a convenience sample of PLWHA in care. Of all the demographic, health status, risk history and behaviors and partnership covariates explored, only the partnership covariates were significantly associated with unprotected intercourse. Significant covariates included having a steady partner (OR and 95%CI = 4.2; 1.3, 13.5), HIV-positive (OR and 95%CI = 2.7; 1.0, 6.9 versus HIV-negative partner) or unknown serostatus partner (OR and 95%CI = 4.6; 1.1, 18.3 versus HIV-negative partner) and men who have sex with men (MSM) partnerships (OR and 95%CI = 3.0; 1.2, 7.3). Partnership covariates explained 23% of the variance in unprotected intercourse; other groups of covariates did not significantly improve model fit. Significant interaction terms between reported partner HIV status, partnership type and sexual orientation revealed the greatest likelihood of unprotected intercourse in two groups of individuals: those in steady relationships with HIV-positive partners and MSM in relationships with partners of unknown serostatus. Prevention interventions for PLWHA should focus on partnership characteristics.