By means of prospective cohort data from the Sydney AIDS Project, we report on 55 homosexual or bisexual men who have become infected with human immunodeficiency virus (HIV), as measured by the development of serum HIV antibodies (seroconversion). We have compared the sexual practices, recreational drug abuse, history of sexually-transmissible diseases, and antecedent immunological findings of the men who seroconverted with those of 588 subjects who persistently remained seronegative in the same time-period. The cumulative incidence rate of HIV infection over the three years of observation was 8.5%. The cumulative incidence rate ranged from less than 1% for the six months before August 1, 1984, to a peak of 5% in the six months before August 1, 1985. Of those subjects for whom we had data for the period of seroconversion, all but two of the subjects who seroconverted admitted to a recognized high-risk sexual practice in the six months before the first visit at which they were found to be seropositive. Univariate analysis found that men who seroconverted were significantly more likely to have had a greater number of recent sexual partners (relative risk per partner, 1.02; P less than 0.001), to have engaged in receptive anal intercourse (incidence rate ratio, 3.1; 95% confidence interval [CI], 1.3-7.6; P = 0.01) and to have used nitrite inhalant (incidence rate ratio, 2.6; 95% CI, 1.2-5.9; P = 0.02) and amphetamine (incidence rate ratio, 4.8; 95% CI, 2.2-10.5; P less than 0.001) drugs. The men who seroconverted were significantly (incidence rate ratio, 2.7; 95% CI, 1.2-6.1; P = 0.014) more likely to have antecedent T-suppressor-cell counts of greater than 800 cells/microL. Factors that retained significance in multivariate analysis were the number of recent sexual partners, recent amphetamine abuse and a T-suppressor-cell count of greater than 800 cells/microL.