Recent evidence of the evolutionary adaptation of HIV-1 to the specific immune system is reviewed. Longitudinal studies of patients show that a neutralizing antibody (NAb) response specific to autologous virus is detectable within 1-2 months of infection and that viral variants resistant to neutralization arise and spread in the viral population within the subsequent three months, and that this general pattern is repeated. There is strong evidence that amino acid replacements in gp120 glycan-binding motifs affect viral sensitivity to neutralization and are selected by NAbs. Longitudinal studies of humans have also provided good evidence of amino acid replacements in cytotoxic T lymphocyte (CTL) epitopes that allow the virus to escape detection by CTLs. But, the clearest evidence of adaptation to CTL surveillance at the molecular level comes from experiments with SIV-infected rhesus macaques. These show unequivocally that amino acid replacements in CTL epitopes are the result of positive selection and that these escape mutants have a lower class I major histocompatibility complex (MHC) binding affinity or are less likely to be recognized by CTLs than non-escape variants. To improve our ability to predict HIV's evolutionary responses to selection by the specific immune system it is suggested that future work focus on the details of the adaptive response to antibody surveillance, the temporal dynamics of specific immune responses, the relative importance of antibody and CTL selection, and the effects of superinfection, viral recombination, and viral protein functional constraints on immune escape.