This paper develops a real options approach to the optimal sequencing of antiretroviral drug cocktails for HIV/AIDS patients in resource-poor settings. The analysis focuses on the implications of endogenous resistance mutations in the virus that reduce or eliminate the effectiveness of individual drugs within a cocktail when lack of laboratory equipment prevents these from being identified. Using a model with two drug cocktails, we show that the first-line therapy should be introduced later than in the case without resistance mutations and that the second-line therapy should be introduced earlier. We go on to discuss implications for comparative cost-effectiveness analyses.