OBJECTIVE: To describe the safety and tolerability of zidovudine, lamivudine, and efavirenz in a low-income setting. DESIGN: We conducted a prospective cohort study in a workplace HAART programme in South Africa, which uses a first-line regimen of efavirenz, zidovudine, and lamivudine and provides routine clinical and laboratory monitoring 6-monthly pre-HAART and at 2, 6, 12, 24, 36, 48 weeks during HAART. METHODS: We assessed the incidence of specified clinical and laboratory events (AIDS Clinical Trials Group grade 3 or higher) and associated regimen changes, hospitalizations, and deaths one year before HAART initiation and one year on-HAART using person-year analysis. RESULTS: Between November 2002 and October 2005, 853 subjects (98% male, median age 40 years, and median CD4 cell count at HAART initiation 186 cells/mul) met enrollment criteria. The incidence of events on-HAART was higher than pre-HAART for neutropenia and nausea/vomiting. Dizziness was common early after HAART initiation (not evaluated pre-HAART). Of those with neutropenia, 88% had no apparent clinical consequences. The incidence of anemia, hepatotoxicity, peripheral neuropathy, and rash was similar or higher pre-HAART than on-HAART. Mean hemoglobin rose during the time on-HAART and was higher at 24 and 48 weeks than at baseline (P < 0.001). DISCUSSION: This regimen was well tolerated with a short-term increase in neutropenia, nausea, and probably neurocerebellar events. Most significantly, in contrast to reports from high-income countries, we observed a long-term improvement in the hemoglobin concentration.