To determine the feasibility and effectiveness of integrating highly active antiretroviral therapy (HAART) into existing tuberculosis directly observed therapy (TB/DOT) programs, we performed a pilot study in an urban TB clinic in South Africa. Patients with smear-positive pulmonary TB were offered HIV counseling and testing. Twenty HIV-positive patients received once-daily didanosine (400 mg) plus lamivudine (300 mg) plus efavirenz (600 mg) administered concomitantly with standard TB therapy Monday to Friday and self-administered on weekends. After completing TB therapy, patients were referred to an HIV clinic for continued treatment. At baseline, patients had a mean CD4 count of 230 cells/mm(3) (range: 24-499 cells/mm(3)) and a mean viral load of 5.75 log(10) (range: 3.81-7.53 log(10)). Seventeen completed combined standard TB and HIV therapy; 16 of 20 (80%) patients enrolled and 15 of 17 (88%) patients completing standard TB therapy achieved a viral load <50 copies/mL and mean CD4 count increase of 148 cells/mm(3). TB was cured in 17 of 20 (85%) enrolled patients and 17 of 19 (89%) patients with drug-sensitive TB. Treatment was well tolerated, with minimal gastrointestinal, hepatic, skin, or neurologic toxicity. The project was well accepted and integrated into the daily TB clinic functions. This pilot study demonstrates that TB/DOT programs can be feasible and effective sites for HIV identification and the introduction and monitoring of a once-daily HAART regimen in resource-limited settings.