The Sydney AIDS Project is a prospective immunoepidemiological study of 996 homosexual/bisexual men enrolled between February 1984 and January 1985. By January 1987, 32 of 386 homosexual men who were seropositive at enrollment in the study had developed AIDS, yielding a crude progression rate of between 2.8% and 4.2% per annum. Of these subjects, 23 (72%) developed AIDS within 12 months of enrollment. In univariate analysis, the only lifestyle differences between seropositive subjects who progressed to AIDS and those that did not progress were less frequent oral sex activity and more use of marijuana in the three months prior to enrollment. In multivariate analysis, seropositive subjects who progressed to AIDS were more likely to have a lower percentage of CD4+ cells, a higher percentage of CD8+ cells and to have used marijuana in the three months prior to enrollment than the seropositive subjects who did not progress. No HIV seropositive subject who was asymptomatic and had normal T-cell subsets at enrollment had developed AIDS by January 1987. Persistent generalised lymphadenopathy was not associated with progression to AIDS. Although there are a number of lifestyle factors that may be associated with HIV infection, this study did not implicate most of these in the progression of HIV seropositive subjects to end-stage AIDS. We conclude that antecedent changes in T-cell subsets are associated with progression to AIDS and we emphasise the prognostic value of enumeration of T-cell subsets in HIV seropositive persons.