We compared the efficacy and the toxicity of zidovudine (AZT) versus stavudine (d4T), in combination with lamivudine (3TC) and indinavir, in AZT-, dideoxyinosine (ddI)-, and/or dideoxycytosine (ddC)-experienced patients in a randomized comparative multicenter trial. One hundred seventy human immunodeficiency virus type 1 (HIV-1)-infected patients, who had received AZT, ddI, and/or ddC for at least 6 months but were naive for d4T, 3TC, and protease inhibitors, were randomized to AZT at 250 to 300 mg twice daily, 3TC at 150 mg twice daily, and indinavir at 800 mg every 8 h or to d4T at 40 mg twice daily, 3TC at 150 mg twice daily, and indinavir at 800 mg every 8 h. The primary endpoint was time to virological failure, defined as plasma HIV-1 RNA levels of >5,000 copies/ml after at least 8 weeks of antiretroviral therapy. Additional endpoints were change from baseline in CD4 cell counts, AIDS-defining events and adverse events, and proportion of patients with HIV-1 RNA levels of <500 copies/ml and HIV-1 RNA levels of <50 copies/ml. At week 80, 15 patients in the AZT arm and 14 patients in the d4T arm had reached the primary endpoint, and time to virological failure did not differ between the two arms (P = 0.98). In the d4T and in the AZT arms, 67 and 73% of patients, respectively, had HIV-1 RNA levels of <500 copies/ml (P = 0.50). The median change from baseline in CD4 cell count was 195 x 10(6) and 175 x 10(6)/liter for the d4T- and AZT-containing arms, respectively. The proportions of patients with HIV-1 RNA levels of <50 copies/ml at weeks 8, 16, and 24 were similar in the two arms. The occurrence of serious adverse events was not significantly different between arms. In conclusion, in these patients heavily pretreated with AZT, switching from AZT to d4T when initiating indinavir and 3TC did not bring any additional benefit compared to maintaining AZT.