Background and objective: Worldwide, 50% of human immunodeficiency virus (HIV)-infected people are women. This study was to evaluate whether the safety and efficacy outcomes of three initial antiretroviral regimens (ARVs) differed by sex. Methods: Antiretroviral regimen naive participants from nine countries in four continents were assigned to ARVs with efavirenz (EFV) plus lamivudine-zidovudine, atazanavir (ATV) plus didanosine (ddI)-EC/emtricitabine (FTC) or EFV plus FTC-tenofovir-DF. The primary objective was to estimate the sex difference on efficacy outcome of treatment failure defined as one of the following: 1. Time to 1st of confirmed virologic failure, 2. WHO Stage 4 progression or 3. death with hazard ratio (HR) and 95% confidence interval (CI) from adjusted Cox regression models. Results: In all, 739 (47%) women and 832 (53%) men with HIV were evaluated. Women had higher pretreatment CD4z(182 vs 165 cells/mm3; Pv0.001) and lower HIV-1 RNA (4.9 log10 vs 5.2 log10 copies/ml; Pv0.001) compared to men. Association of sex with time to regimen failure differed by treatment arm (P50.018). For atazanavir plus didanosine-EC plus emtricitabine, women had a longer time to treatment failure compared to men [adjusted HR (aHR)50.59; 95% CI 0.40-0.87]. Women were less likely to prematurely discontinue treatment prematurely (aHR50.74; 95% CI 0.56-0.98). Women assigned to efavirenz plus lamivudine-zidovudine were more likely to have a primary safety event compared to men (aHR51.49; 95% CI 1.18-1.88). Conclusion: Antiretroviral efficacy and safety differed by sex in this study. Consideration of potential effects of sex on antiretroviral outcomes is important for the design of future clinical trials and for HIV treatment guidelines.