A number of obstacles remain in the search for an animal model for HIV infection and pathogenesis that can serve to predict efficacy in humans. HIV-1 fails to replicate and cause disease except in humans or chimpanzees, thereby limiting our ability to evaluate compounds or vaccines prior to human testing. Despite this limitation, nonhuman primate lentivirus models have been established that recapitulate the modes of infection, disease course, and antiviral immunity that is seen in HIV infection of humans. These models have been utilized to understand key aspects of disease and to evaluate concepts in therapies and vaccine development. By necessity, animal models can only be validated after successful trials in humans and the determination of correlates of protection. Because the only vaccine product tested in phase III trials in humans failed to achieve the desired protective threshold, we are as yet unable to validate any of the currently used nonhuman primate models for vaccine research. In the absence of a validated model, many experts in the field have concluded that prophylactic vaccines and therapeutic concepts should bypass primate models, and rely solely upon the systematic testing of each individual and combined vaccine element in human phase I or I/II trials to determine their relative merits. Indeed, a large effort is underway to expand efforts to test all products as part of an international effort termed "The HIV Vaccine Enterprise", with major contributions from the Bill and Melinda Gates Foundation. This Herculean task could potentially be reduced if it were possible to utilize even partially validated nonhuman primate models as part of the screening efforts. The purpose of this article is to review the data from nonhuman primate models that have contributed to our understanding of lentivirus infection and pathogenesis, and to critically evaluate how well these models have predicted outcomes in humans. Key features of the models developed to date are described and their contributions to HIV pathogenesis, therapeutics, and vaccines, are compared. This analysis shows that many of the models at hand have yielded data on drug action and immune responses to vaccines that are congruent with clinical data. This finding suggests that primate models are valuable as adjunctive testing systems to prioritize future therapeutic and vaccine strategies. Nonhuman primate testing of vaccine approaches in particular has provided valuable information and can significantly enhance and accelerate the evaluation of novel concepts necessary to achieve acceptable levels of efficacy. Because major gaps remain in the quest for fully effective vaccines and therapies, it seems prudent to continue aggressive research programs in the nonhuman primate models.