BACKGROUND Perceptions of personal efficacy have been consistently identified as being determinants of exercise adherence in asymptomatic, rehabilitative, younger, and older populations. The present study incorporated a randomized control design in an effort to examine the effects of an efficacy-based intervention in enhancing exercise adherence in a large sample (N = 114) of formerly sedentary middle-aged males and females. METHODS Subjects randomly assigned to an exercise plus intervention group or an exercise plus attentional control group participated in a 5-month long walking program led by trained personnel. Exercise behavior (frequency, miles walked, duration) were assessed on a continuous basis and self-efficacy was measured at 1, 2, and 4 months. RESULTS Repeated measures multivariate analyses revealed a significant treatment effect with subjects in the intervention group exercising more frequently, for longer duration, and walking greater distances over the course of the program. Path analysis indicated that the effect of the treatment on adherence was direct rather than through self-efficacy as hypothesized. Self-efficacy was, however, a significant predictor of exercise behavior in the early and middle stages of the exercise program but not during the last month. CONCLUSIONS An intervention program designed to maximize information pertaining to participants' capabilities appears to have had a reasonable effect on reducing attrition in middle-aged males and females and self-efficacy was a significant predictor of exercise frequency over time. Further research efforts are required to tease out those cognitive factors that might underlie any effects of interventions in exercise adherence.