BACKGROUND Stavudine is a promising antiretroviral agent, but its clinical efficacy has not been determined. OBJECTIVE To evaluate the clinical effect of stavudine (2',3'-didehydro-3'-deoxythymidine) monotherapy in patients with human immunodeficiency virus (HIV) infection. METHODS Randomized, controlled, double-blind trial. METHODS 56 outpatient clinics in private practices, universities, and contract research organizations in the United States, France, and Italy. METHODS 822 HIV-infected adults who had 50 to 500 CD4+ cells/mm3 and had previously received at least 6 months of zidovudine treatment. METHODS Monotherapy with peroral stavudine capsules or peroral zidovudine capsules. METHODS The primary end point was clinical progression, which was defined as all occurrences of acquired immunodeficiency syndrome (AIDS)-defining events or death. RESULTS Patients receiving stavudine reached clinical end points at a rate of 26 per 100 person-years, compared with 32 per 100 person-years for patients receiving zidovudine (relative risk, 0.75 [95% CI, 0.58 to 0.98]; P = 0.03). The risk for death alone was 26% lower in the stavudine group than in the zidovudine group, but the comparison was not statistically significant (relative risk, 0.74 [CI, 0.53 to 1.02]; P = 0.066). The benefit of stavudine therapy was seen in all CD4+ cell strata (< or = 100 cells/mm3, 101 to 300 cells/mm3, and > 300 cells/mm3) and clinical stages of HIV disease (asymptomatic, symptomatic, and AIDS). Four weeks after treatment began, CD4+ cell counts were 30 cells/mm3 higher in the stavudine group than in the zidovudine group; this difference was sustained for 96 weeks (P < 0.001). Nausea and vomiting were more common in patients receiving zidovudine (P < 0.01), and neuropathy occurred more frequently in those receiving stavudine (12% in the stavudine group compared with 4% in the zidovudine group; P < 0.001). Neuropathy resolved completely in many patients (63%) after interruption of stavudine treatment; these patients could resume stavudine therapy at a lower dose. CONCLUSIONS Stavudine was well tolerated and delayed progression of HIV disease in patients who had previously received 6 or more months of zidovudine treatment. Benefits were apparent in all CD4+ cell strata and clinical stages of HIV disease. Stavudine is an important agent to consider for trials of combination chemotherapy.