OBJECTIVE: To identify three groups of obese people entering weight loss treatment, who have distinctly different cognitive appraisals of dietary transgressions and to compare these groups on self-report inventories of eating patterns, dieting, and depression, as well as on treatment completion rates and weight loss. DESIGN: Retrospective review of clinical records. Using a measure which evaluates eating-related cognitive appraisals, participants were categorized into one of three cognitive groups (All-or-None, Rationalization, Matterof-Degree). SUBJECTS: 289 treatment-seeking obese women (age: 40.9 y, body mass index (BMI): 34.7 kgam 2 ). MEASUREMENTS: Self-reported eating and dieting behavior (Three-Factor Eating Questionnaire and Eating Behavior Inventory); depression (Beck Depression Inventory); attendance information and body weight obtained during treatment. RESULTS: The cognitive group representing objective thinkers (Matter-of-Degree) reported signi®cantly fewer problems with overeating and more personal control over eating than did the rigid, dichotomous thinkers (All-orNone). In addition, the Matter-of-Degree (MAT) group endorsed signi®cantly less subjective hunger and fewer depressive symptoms than the other two cognitive groups. The Rationalization group was more likely to complete a treatment program than was the All-or-None group, with the MAT group not differing from either. Despite these ®ndings, there were no signi®cant differences among cognitive groups on total weight loss. CONCLUSIONS: Cognitive appraisals of weight-control lapses appear to be associated with self-reported eating behavior, depressive symptoms and treatment completion rates, but not with treatment-induced weight loss. The relationship between long-term weight loss and cognitive appraisals of dieting lapses is yet to be determined. It appears necessary to assess empirically the validity of assumptions regarding factors associated with treatment outcome.