Architects of behavioral weight-loss programs require improved understandings of psychosocial correlates of healthy eating behaviors to improve overwhelmingly poor outcomes. Exercise might benefit eating changes through a transfer of self-regulatory skill improvements. However, related research is minimal, and degrees of obesity might influence the utilization and efficacy of self-regulatory processes. Women of Class 1 obesity (body mass index [BMI] = 30–34.9 kg/m 2 ; n = 53), Class 2 obesity (BMI = 35–39.9 kg/m 2 ; n = 41), and Class 3 obesity (BMI ≥ 40 kg/m 2 ; n = 53) enrolled in a community-based cognitive-behavioral weight-management program were assessed over 6 months on changes in exercise, self-regulation for both exercise and eating, and the consumption of fruits/vegetables and sweets. Within-group improvements were all significant, with the greater self-regulation changes consistent with lower degrees of obesity. Within serial multiple mediation analyses, the hypothesized path from exercise change→self-regulation for exercise change→self-regulation for eating change→eating behavior change was supported in both the prediction of changes in fruit/vegetable intake and sweets consumption. However, the direct effect of increased exercise on improved eating was not significant. BMI score or obesity class did not significantly moderate embedded relationships among variables. Findings suggested weight-loss programs incorporate supported exercise, and dedicate high proportions of their curricula to the development of self-regulatory skills.