Objective: To determine the effectiveness of changes in factors associated with self-efficacy theory for predicting weight change in obese women of two ethnic groups. Design: Obese (body mass index [BMI] ≥30 kg/m2) White (n=34) and African American (n=30) women (mean age 44 years) were assessed on measures of body satisfaction, self-efficacy, and weight change over 20 weeks. Setting: Community wellness centers. Intervention: A supported exercise and nutrition information treatment. Main outcome measures: Changes in Body Areas Satisfaction Scale (BAS), Physical Self-Concept Scale (PSCS), and Exercise Self-Efficacy Scale (ESES) scores and changes in body weight over 20 weeks. Results: Significant improvements were found on measures of exercise self-efficacy, body satisfaction, and weight, with a trend (P=.073) toward significantly greater mean weight loss by the White group (-15.5 kg vs -9.1 kg). Linear multiple regression analyses, with simultaneous entry of changes in BAS, PSCS, and ESES scores, significantly predicted changes in weight for both the White (R2=.25) and African American (R2=.50) group. The primary predictor of weight change for the White group was change in BAS scores (β=-.42) and for the African American group was change in ESES scores (β=-.68). Conclusions: Self-efficacy theory was supported as an explanatory model for both groups, with notable differences. Implications for weight loss intervention design and application are discussed.