BACKGROUND The efficacy and safety of adding a protease inhibitor to two nucleoside analogues to treat human immunodeficiency virus type 1 (HIV-1) infection are not clear. We compared treatment with the protease inhibitor indinavir in addition to zidovudine and lamivudine with treatment with the two nucleosides alone in HIV-infected adults previously treated with zidovudine. METHODS A total of 1156 patients not previously treated with lamivudine or protease inhibitors were stratified according to CD4 cell count (50 or fewer vs. 51 to 200 cells per cubic millimeter) and randomly assigned to one of two daily regimens: 600 mg of zidovudine (or stavudine) and 300 mg of lamivudine, or that regimen with 2400 mg of indinavir. The primary end point was the time to the development of the acquired immunodeficiency syndrome (AIDS) or death. RESULTS The proportion of patients whose disease progressed to AIDS or death was lower with indinavir, zidovudine, and lamivudine (6 percent) than with zidovudine and lamivudine alone (11 percent; estimated hazard ratio, 0.50; 95 percent confidence interval, 0.33 to 0.76; P=0.001). Mortality in the two groups was 1.4 percent and 3.1 percent, respectively (estimated hazard ratio, 0.43; 95 percent confidence interval, 0.19 to 0.99; P=0.04). The effects of treatment were similar in both CD4 cell strata. The responses of CD4 cells and plasma HIV-1 RNA paralleled the clinical results. CONCLUSIONS Treatment with indinavir, zidovudine, and lamivudine as compared with zidovudine and lamivudine alone significantly slows the progression of HIV-1 disease in patients with 200 CD4 cells or fewer per cubic millimeter and prior exposure to zidovudine.