Purpose: This study aimed at assessing whether psychosocial predictors of controlled eating and weight loss also predict emotional eating, and how differing weight-loss treatment methods affect those variables. Methods: Women with obesity (M = 47.8 ± 7.9 years; BMI = 35.4 ± 3.3 kg/m2) were randomized into groups of either phone-supported self-help (Self-Help; n = 50) or in-person contact (Personal Contact; n = 53) intended to increase exercise, improve eating behaviors, and reduce weight over 6 months. Results: A multiple regression analysis indicated that at baseline mood, self-regulating eating, body satisfaction, and eating-related self-efficacy significantly predicted emotional eating (R2 = 0.35), with mood and self-efficacy as independent predictors. Improvements over 6 months on each psychosocial measure were significantly greater in the Personal Contact group. Changes in mood, self-regulation, body satisfaction, and self-efficacy significantly predicted emotional eating change (R2 = 0.38), with all variables except self-regulation change being an independent predictor. Decreased emotional eating was significantly associated with weight loss. Conclusion: Findings suggest that weight-loss interventions should target specific psychosocial factors to improve emotional eating. The administration of cognitive-behavioral methods through personal contact might be more beneficial for those improvements than self-help formats.