The purpose of the study is to evaluate the performance of theoretically-derived mediators of health behavior change. Participants were university seniors (184 females; 154 males) randomly assigned to an intervention course designed to promote physical activity or to a control course. Five physical activity outcomes and nine psychosocial mediating variables were assessed at baseline and the end of the 16-week course. For women, the intervention had significant effects on five of the mediators, including self-efficacy for making time, self-efficacy for resisting relapse, social support from friends, and experiential and behavioral processes of change. Among men, the intervention improved use of behavioral processes of change but also had the unintended effect of increasing perceived barriers to activity. For women, significant contributors to regressions explaining physical activity change were social support from friends (for total activity) and change in self-efficacy for resisting relapse (for vigorous exercise). For men, significant explanatory variables included change in enjoyment (for total activity), change in self-efficacy for resisting relapse (for strength exercise), and change in benefits (for moderate intensity activity). For both sexes, there were significant findings in the unexpected direction. Across the five physical activity outcomes, hypothesized mediators were inconsistent and weak contributors to the models. Investigating mediators of behavior change has the potential to stimulate improvements in theories and interventions.