In a group of homosexual men in Vancouver studied prospectively since November 1982, 26 cases of acquired immune deficiency syndrome (AIDS) have arisen. To identify behavioural, clinical and laboratory findings that might predict the development of AIDS in people with antibody to human immunodeficiency virus (HIV), we compared data for 25 patients with AIDS with corresponding data for 80 controls serologically positive for HIV selected from the cohort. The clinical and laboratory data for the patients with AIDS preceded the diagnosis of the syndrome by a mean of 17.5 months. The controls had been both seropositive and AIDS-free for a mean of 16.7 months after acquisition of their data. We detected significant differences between the patients with AIDS and the controls in IgG and IgA levels, absolute number of helper T cells and ratio of helper to suppressor T cells but not in lifetime number of male sexual partners, frequency of receptive anal intercourse or receptive fisting, illicit drug use or history of infectious disease. We also detected an increased risk of AIDS among those who had an elevated number of sexual contacts in AIDS-endemic areas in the 5 years before enrollment. A history of increased early sexual contact in AIDS-endemic areas is likely to be associated with early infection and with an increased risk of AIDS among men with HIV infection of unknown duration. Thus, although our analysis had limited statistical power, we conclude that most lifestyle variables appear to act as exposure factors in HIV infection but not as cofactors in the development of AIDS.