Data for this study came from the Urban Men's Health Study, a telephone probability sample of men who have sex with men in four U.S. cities (San Francisco, New York, Los Angeles, and Chicago) conducted between November 1996 and February 1998. The process of screening households for MSM eligibility was performed exclusively by male interviewers. A total of 2,881 complete interviews were obtained, a participation rate of 78 percent of all MSM eligible households. The scope of the survey instrument permitted the identification of indicators of many possibly pertinent long-term sequelae of sexual abuse, e.g., self-destructive behavior, problematic substance use, and disturbed/distorted relationship patterns. The men were asked about various types of sexual behavior and whether protective measures were used to reduce the risk of HIV infection. Independent variables included demographics, whether the interviewee was sexually abused as a child, adverse family experiences, substance use, depression, abusive relationships, "one-night stands," and adult sexual revictimization. The presence of CSA among the sample was estimated to be 20.6 percent, and another 14.7 percent reported sexual victimization as an adult (age 18 or older), and 64.7 percent reported no sexual coercion experiences. On average, CSA victims were preteens at the time of their first CSA experience, and perpetrators were approximately 16 years older than their victims. The perpetrators were typically male and not likely to be family members. Initial CSA experiences were characterized by high levels of force; 43 percent involved physical force/weapons, and penetrative sex (78 percent; 46 percent reported attempted or actual anal intercourse). The men who were forced into sexual acts as children were more likely than never-coerced men to engage in high-risk sexual behaviors (unprotected anal intercourse with a nonprimary partner or with a serodiscordant male). The multivariate analyses showed that the effect of childhood sexual coercion on sexually risky behaviors was mediated by substance use, patterns of sexual contacts, and partner violence, but not by adult sexual revictimization or by depression. These findings are interpreted within the context of social learning theory and prior research on sexual risk-taking. This study, along with earlier studies, suggests the importance of considering childhood sexual abuse history in developing and delivering HIV interventions. Suggestions for further research are offered. 6 tables and 79 references