BACKGROUND: The optimal sequencing of antiretroviral regimens for the treatment of infection with human immunodeficiency virus type 1 (HIV-1) is unknown. We compared several different antiretroviral treatment strategies. METHODS: This multicenter, randomized, partially double-blind trial used a factorial design to compare pairs of sequential three-drug regimens, starting with a regimen including zidovudine and lamivudine or a regimen including didanosine and stavudine in combination with either nelfinavir or efavirenz. The primary end point was the length of time to the failure of the second three-drug regimen. RESULTS: A total of 620 subjects who had not previously received antiretroviral therapy were followed for a median of 2.3 years. Starting with a three-drug regimen containing efavirenz combined with zidovudine and lamivudine (but not efavirenz combined with didanosine and stavudine) appeared to delay the failure of the second regimen, as compared with starting with a regimen containing nelfinavir (hazard ratio for failure of the second regimen, 0.71; 95 percent confidence interval, 0.48 to 1.06), as well as to delay the second virologic failure (hazard ratio, 0.56; 95 percent confidence interval, 0.29 to 1.09), and significantly delayed the failure of the first regimen (hazard ratio, 0.39) and the first virologic failure (hazard ratio, 0.34). Starting with zidovudine and lamivudine combined with efavirenz (but not zidovudine and lamivudine combined with nelfinavir) appeared to delay the failure of the second regimen, as compared with starting with didanosine and stavudine (hazard ratio, 0.68), and significantly delayed both the first and the second virologic failures (hazard ratio for the first virologic failure, 0.39; hazard ratio for the second virologic failure, 0.47), as well as the failure of the first regimen (hazard ratio, 0.35). The initial use of zidovudine, lamivudine, and efavirenz resulted in a shorter time to viral suppression. CONCLUSIONS: The efficacy of antiretroviral drugs depends on how they are combined. The combination of zidovudine, lamivudine, and efavirenz is superior to the other antiretroviral regimens used as initial therapy in this study.