BACKGROUND: Expanded access to antiretroviral therapy (ART) has produced dramatic reductions in HIV-associated morbidity and mortality. Disparities in access to and benefit from ART have been observed by race, gender, and mental health status, however. METHODS: From 2001 to 2002, 611 HIV-positive patients were consecutively recruited from 5 southeastern US states and followed for 3 years. We evaluated demographic and psychosocial predictors of probability of receiving ART among all those eligible for ART (on ART, CD4 <350 cells/mm or viral load [VL] >55,000 copies/mL in the year preceding enrollment), time to first ART discontinuation among those on ART, and time to VL >400 copies/mL among those on ART with VL <400 copies/mL at enrollment. RESULTS: Of 611 participants, 474 consented to medical record abstraction and had known ART status at enrollment; 81% (385 of 474) of all participants and 89% (385 of 435) of ART-eligible participants were receiving ART at enrollment. In multivariable analyses, ART receipt was associated with greater age (adjusted odds ratio = 1.92 per 10 years, 95% confidence interval: 1.23 to 3.01), fewer recent stressful life events (odds ratio = 0.68, 95% confidence interval: 0.51 to 0.92), less alcohol use (odds ratio = 0.64, 95% confidence interval: 0.46 to 0.90), and greater perceived self-efficacy (OR = 2.82, 95% confidence interval: 1.41 to 5.62). No psychosocial characteristics were associated with ART discontinuation or virologic failure. No racial/ethnic or gender disparities were observed in ART receipt; however, minority racial/ethnic groups were faster to discontinue ART (adjusted hazard ratio = 2.44, 95% confidence interval: 1.33 to 4.49) and experience virologic failure (adjusted hazard ratio = 2.01, 95% confidence interval: 1.09 to 3.71). CONCLUSIONS: Patients with unfavorable psychosocial profiles were less likely to be on ART, perhaps attributable to providers' or patients' expectations of readiness. Psychosocial characteristics were not associated with ART discontinuation or virologic failure, however, possibly reflecting the selection process involved in who initiates ART. Racial disparities in ART discontinuation and virologic failure merit further attention. © 2008 Lippincott Williams & Wilkins, Inc.