The National Institute of Mental Health Multisite HIV Prevention Trial was a trial of an intervention to reduce sexual HIV risk behaviors among 3,706 low-income at-risk men and women at 7 U.S. research sites. The intervention, based on social-cognitive theory and designed to influence behavior change by improving expected outcomes of condom use and increasing knowledge, skills, and self-efficacy to execute safer sex behaviors, was effective relative to a control condition in reducing sexual risk behavior. At 3 months after completion of the intervention, measures of these potential mediators were higher in the intervention than in the control condition. Although the effect of the intervention on sexual risk behavior was significantly reduced when the variables were controlled statistically, supporting the hypothesis of their mediation of the intervention effect, most of the effect remained unexplained, indicating the influence of unmeasured factors on outcome.