Despite considerable self-initiated HIV risk reduction among men who have sex with men, little is known about how to design interventions that will effectively assist individuals from this population in maintaining safer sex behaviors over time. The present study evaluated the effectiveness of a 17-session group counseling intervention that incorporated components based on a cognitive-behavioral model of relapse. Differential behavioral outcomes following treatment included an increase in the percentage of sexual activities that were protected and a decrease in unprotected oral sex. However, considerable risk reduction (e.g., increased condom use, decreased unprotected sex, and decreases in the number of male partners and in the total number of sexual acts) occurred in both treated and untreated participants. Measures of mediating attitudinal variables drawn from relapse prevention theory largely predicted behavioral changes. Over time, several of the risk reduction behaviors achieved at posttreatment were not maintained, suggesting the importance of further developing effective strategies for supporting behavior change maintenance.