Predictors of unprotected anal intercourse were examined among 508 gay men in San Francisco. The cohort was recruited in 1983-84 at which time 49.8 percent of non-monogamous men (N = 435) and 71.2 percent of monogamous men (N = 73) reported practicing unprotected anal intercourse. Only 12 percent of non-monogamous and 27.4 percent of monogamous men reported these practices in 1988. The non-monogamous men who practiced unprotected anal intercourse in 1984 were more likely to be younger, to report that unprotected anal intercourse was their favorite sexual activity, to be low in perceived efficacy to change sexual behavior, to report that friends were more likely to engage in high-risk behaviors, to have less knowledge of health guidelines, and to be less depressed at that time. Non-monogamous individuals who in 1984 reported that unprotected anal intercourse was their favorite sexual activity were more likely to practice that behavior in 1988. Those who knew their serostatus as positive were less likely to report unprotected anal intercourse in 1988. These data infer that in order to modify AIDS-related high-risk behaviors, community risk-reduction programs be differentially aimed at young persons so as to increase personal efficacy about risk reduction, challenge peer norms, promote antibody testing, and eroticize safer sexual activities.