Selecting the primary endpoint is one of the most important decisions in designing clinical trials. Many HIV trials are powered for surrogate markers, often virologic suppression. Among 49 recently published Phase 3 or higher randomized HIV trials only 14% were powered for clinical outcomes such as the progression to AIDS, death, or serious non-AIDS diseases. We provide two examples where interventions modified the targeted surrogate markers but failed to provide clinical benefit. We review the use of surrogate and clinical endpoints, discuss the composition of clinical endpoints, and the need for endpoint verification. In HIV-infected individuals with CD4 cell counts above 200 cells/mm(3) serious non-AIDS conditions such as cardiovascular, renal, hepatic diseases and cancer contribute substantially to morbidity and mortality. In this population clinical endpoint trials should be powered for non-AIDS morbidity along with AIDS.