Owing to their low toxicity, low price, and ease of use, efavirenz (EFV) and nevirapine (NVP) are frequently used as part of antiretroviral regimens for AIDS treatment. Several clinical trials have already studied their efficacy and tolerance. However, long-term observations of the effects of these drugs in patients are limited. We used data from a prospective Senegalese cohort to analyze long-term tolerance and efficacy of these two drugs in a low-resources setting. Patients were included if they started their therapy with EFV or NVP. They were censored after treatment discontinuation. The primary endpoint was the time to treatment discontinuation. Secondary endpoints included time to death, time to disease progression, occurrence of severe adverse effects, CD4 cell recovery, and virological response. Confounding factors were controlled using marginal structural models. The median follow-up time in both EFV and NVP arms was 48 months. The hazard ratio (HR) of drug discontinuation in the EFV arm vs. the NVP arm was 0.84 (0.34; 1.87). There was a borderline difference in virological response [HR 1.38 (0.999; 1.89)] but no differences in time to death [HR 1.15 (0.41; 3.24)], time to AIDS progression [HR 1.25 (0.61; 2.58)], or time to increase in CD4 cell count above 500 cells/mm3. Adverse effects were different between NVP and EFV, but long-term tolerance was good for both. This analysis provided further information on long-term tolerance and efficacy of EFV and NVP in a resource-limited setting.