BACKGROUND: Obesity is a prominent modifiable health risk factor. Treatments of severe caloric restrictions and educational interventions have had minimal sustained effects on weight loss. Physical activity may have significant indirect effects on weight reduction associated with changes in psychological variables, although explanatory models are lacking. METHODS: Relationships based on Baker and Brownell's model of exercise, self-appraisal, mood change, and weight loss were tested with obese (body mass index ≥30) women initiating a supported exercise and nutrition information program over 6 months. RESULTS: Exercise participation was associated with significant improvements in mood, body image, and exercise-related self-efficacy. When changes on measures of these factors were simultaneously entered into a multiple regression equation, a significant portion of the variance in exercise session attendance was accounted for (R = 0.26, F(7, 52) = 2.57, P < 0.05), with changes in tension (β = -0.34) and physical self-concept (β = 0.33) making significant unique contributions. Exercise session attendance was significantly correlated with weight and body composition changes (r = -0.30 to -0.47). The indirect effect of exercise on weight loss was estimated at 0.23. As hypothesized, less improvement in depression was significantly associated with less improvement in weight and body composition (r values = 0.23 to 0.29). CONCLUSIONS: Physical activity and exercise may have positive effects on sustained weight loss due to associated changes in self-appraisal and mood factors. Early incorporation of moderate exercise into weight management treatments may have considerable value beyond just energy expenditure. Continued testing of explanatory models is warranted. © Copyright 2008 Southern Society for Clinical Investigation.