We compared the tolerability and effectiveness of two major first-line regimens used in resource-limited settings, namely zidovudine-lamivudine-nevirapine and stavudine-lamivudine-nevirapine. HIV-1-infected adults in Cameroon were enrolled in a prospective cohort study between 2001 and 2003. They were eligible if they had AIDS or a CD4 cell count below 350/mm(3), a Karnofsky score over 50%, and no contraindications to antiretroviral treatment. The patients were followed up to 2 years. Of 169 patients, 85 received zidovudine-lamivudine-nevirapine and 84 stavudine-lamivudine-nevirapine. The incidence rates of treatment changes, death, drug resistance, and severe adverse effects were, respectively, 12.0 [95% confidence interval (CI) 7.2-19.9] and 10.9 (CI 6.4-18.3) per 100 person-years; 5.7 (CI 2.8-11.4) and 7.6 (CI 4.2-13.7); 2.9 (CI 1.1-7.7) and 5.0 (CI 2.4-10.6); and 41.7 (CI 30.2-57.6) and 49.1 (CI 36.1-66.6). The Kaplan-Meier curves for the likelihood of remaining on the initial regimen (p = 0.8) and for survival (p = 0.5) did not differ significantly between the groups. In Cox multivariate analysis only a lower baseline CD4 cell count was associated with death (p < 0.001). The proportion of patients with undetectable viral load and the increase in the CD4 cell count were similar in the two groups. Anemia was rare (4% vs. 6%). Five cases of severe peripheral neuropathy and one case of lipodystrophy occurred. This study suggests that the zidovudine-lamivudine-nevirapine combination is a safe first-line treatment, even in settings with few laboratory resources. In view of stavudine toxicity, these results support recommendations calling for a gradual switch from stavudine- to zidovudine-based regimens.