We aimed to examine moderators and mediators of behaviour change in a cognitive lifestyle program for drug-treated overweight hypertensives in Perth, Australia. We collected data at baseline, 4 months (post-intervention) and 1-year follow-up in a randomized controlled trial of a program that focused on weight loss, diet, and exercise. Mediation analysis used regression models that estimate indirect effects with bootstrapped confidence limits. Outcomes examined were saturated fat intake (% energy) and physical activity (hours per week). In total, 90/118 individuals randomized to usual care and 102/123 to the program-completed follow-up. Sex was a moderator of response post-intervention for diet and physical activity, with a greater response among women with usual care and among men with the program. Change in self-efficacy was a mediator of dietary change post-intervention [effect size (ES) -0.055, 95% confidence interval (CI) -0.125, -0.005] and at follow-up (ES 0.054, 95% CI -0.127, -0.005), and in physical activity post-intervention (ES 0.059, 95% CI 0.003, 0.147). These findings highlight different responses of men and women to the program, and the importance of self-efficacy as a mediator. Mediators for physical activity in the longer term should be investigated in other models, with appropriate cognitive measurements, in future trials.