Identifying a relationship between depression and sexual risk behavior in HIV-infected patients could establish a mechanism to enhance prevention efforts. We conducted a cross-sectional analysis using data from the University of Pennsylvania Center for AIDS Research and used ordinal logistic regression to measure the association between depression and non-condom use. 716 men who have sex with men (MSM), 262 heterosexual men and 277 heterosexual women were included. The association between depression and non-condom use was strongest in heterosexual men with and without HIV-infected regular partners (OR 8.53, 95% CI 1.18-61.89 and OR 2.30, 95% CI 0.99-5.36 respectively), but absent in heterosexual women regardless of partner. Although the OR was low in MSM overall, an association was detected in MSM without HIV-infected regular partners (OR 2.44, 95% CI 1.39-4.31). In conclusion, we demonstrated an association between depression and non-condom use driven by heterosexual men and MSM without HIV-infected regular partners. Sexual risk should be addressed when intervening on depressive symptoms in these subgroups.