Background: Although mechanisms are unclear, there is a robust association between exercise and mood improvements. However, beyond ~ 6 months of expected weight loss, weight-management behaviors such as exercise wane as weight gradually regains in most adults. The amount of exercise required to maintain mood improvement is unknown, as is the possible role of theory-based psychosocial changes associated with treatment such as in self-efficacy. Methods: Women who volunteered for a community-based cognitive-behavioral obesity treatment emphasizing exercise and building self-efficacy via self-regulating through lifestyle challenges/barriers, and were of the 86% who reduced their negative mood during its initial 6 months, were allocated into groups who either maintained their negative mood reduction (Sustain group, n = 43) or reverted toward initial levels of negative mood (Revert group, n = 73) during months 6–24. A binary stepwise logistic regression analysis assessed whether exercise amount and/or change in self-efficacy significantly classified Sustain vs. Revert group membership. Results: Change in exercise amount was not significantly associated with group membership. However, after additional entry of change in exercise self-efficacy, a significant association was found — successfully classifying 70% and 42% of the Revert and Sustain group participants, respectively. When self-efficacy change was entered as the sole independent variable, group membership was classified with 62% accuracy, overall (96% in the Revert group). Conclusions: Consistent with behavioral explanations of the exercise-mood change relationship, it was interpreted that self-efficacy, rather than exercise amount, was an important correlate of mood change beyond month 6. Suggestions for bolstering self-efficacy later in behavioral obesity treatments were provided. [ABSTRACT FROM AUTHOR]