Scaling up access to highly active antiretroviral therapy (HAART) requires eligibility criteria that safeguard treatment efficiency in resource-poor settings. We determined whether supply of HAART on the basis of plasma viral load testing could result in a stronger reduction of AIDS incidence as compared with CD4 count-driven strategies. Expected AIDS incidence rates corresponding to distinct HAART eligibility criteria were calculated by relying on risk parameters obtained through the Amsterdam cohort studies on HIV infection and AIDS. We modeled 2 different treatment settings derived from sub-Saharan African surveys. In a hospital-based setting, the reduction in the 1-year AIDS incidence is the same for any HAART administration rate if patients are selected on a single CD4 cell count criterion or on (additional) criteria for plasma HIV-1 RNA. In a community-based setting, where patients are identified at less advanced stages of infection, the reduction in the 1-year AIDS incidence is higher at particular HAART administration rates if patients are selected on criteria for plasma HIV-1 RNA rather than CD4 cell count. Plasma viral load testing can ensure a more efficient allocation of antiretroviral therapy but only when applied to a strategy of active case finding in the community.