In the absence of direct epidemiological evidence, molecular evolutionary studies of primate lentiviruses provide the most definitive information about the origins of human immunodeficiency virus (HIV)–1 and HIV–2. Related lentiviruses have been found infecting numerous species of primates in sub–Saharan Africa. The only species naturally infected with viruses closely related to HIV–2 is the sooty mangabey (Cercocebus atys) from western Africa, the region where HIV–2 is known to be endemic. Similarly, the only viruses very closely related to HIV–1 have been isolated from chimpanzees (Pan troglodytes), and in particular those from western equatorial Africa, again coinciding with the region that appears to be the hearth of the HIV–1 pandemic. HIV–1 and HIV–2 have each arisen several times: in the case of HIV–1, the three groups (M, N and O) are the result of independent cross–species transmission events. Consistent with the phylogenetic position of a ‘fossil’ virus from 1959, molecular clock analyses using realistic models of HIV–1 sequence evolution place the last common ancestor of the M group prior to 1940, and several lines of evidence indicate that the jump from chimpanzees to humans occurred before then. Both the inferred geographical origin of HIV–1 and the timing of the cross–species transmission are inconsistent with the suggestion that oral polio vaccines, putatively contaminated with viruses from chimpanzees in eastern equatorial Africa in the late 1950s, could be responsible for the origin of acquired immune deficiency syndrome.