This randomized trial examined the effects of 2 oncologist-based exercise interventions--recommendation only (RO) and recommendation plus referral (RR)--versus usual care (UC) on social-cognitive constructs from the theory of planned behavior (TPB). The authors also examined whether the TPB mediated the significant effect of the RO intervention on exercise and explained the null effect of the RR intervention. Independent t tests revealed that both interventions had significant effects on TPB constructs; however, only the RO intervention impacted perceived behavioral control (PBC). Path analyses indicated that PBC was the only construct with a direct effect on exercise and that it mediated the effect of the RO intervention on exercise and explained the null effect of the RR intervention.