It is likely that a successful vaccine against HIV will need to stimulate the innate immune system, generate high levels of neutralising antibody, strong cellular immune responses, and mucosal immunity. Early efforts to develop HIV vaccines attempted to use the virus glycoprotein, gp120, to induce neutralising antibody, but did not take into account the trimeric structure of the native glycoprotein or the complex nature of the CD4 and chemokine receptor binding sites. Recently, attention has been focused on cellular immune responses, particularly T-cell cytotoxicity, based on evidence from the SIV model and from exposed and uninfected humans. Recent experiments in macaques and man suggest that a prime boost regimen using DNA and recombinant pox virus is highly effective at stimulating cellular immunity. However, in addition to the problems of generating neutralising antibodies and mucosal immunity, the difficulty of inducing broad cellular responses able to protect against all clades of HIV, remains an important issue.