The effectiveness of cognitive-behavior treatment (CBT) in nonalcoholic fatty liver disease (NAFLD), largely related to overweight/obesity and considered the hepatic expression of the metabolic syndrome (MS), has so far been tested in very limited samples. In a tertiary referral center, consecutively observed NAFLD subjects were offered a CBT program aimed at weight loss and increased physical activity, based on 13 group sessions; 68 cases entered the treatment protocol, those who refused (n = 82) were given recommendations for diet and physical activity. Treatment goals (weight loss ≥7% initial body weight, normalization of liver enzymes, and improved parameters of MS) were tested by logistic regression at 6 months (all cases) and at 2 years, both on intention-to-treat and in completers (Diet, 78; CBT, 65). The results were adjusted for the propensity score of attending the CBT program, based on civil, anthropometric and clinical variables. At baseline the CBT group had a larger prevalence of obesity and more severe insulin resistance (homeostasis model assessment (HOMA)). At follow-up, CBT was associated with a higher probability of weight loss and normal liver enzymes (6-month: odds ratio (OR), 2.56; 95% confidence interval (CI), 1.15–5.69; 2-year intention-to-treat: OR, 3.57, 95% CI, 1.59–8.00), after adjustment for propensity and changes in body weight. A similar trend was observed in the outcome goals of insulin resistance and the score of MS, which were both reduced. In conclusion, subjects with NAFLD participating in a CBT program significantly improve their general and liver parameters. The beneficial effects are largely maintained at 2-year follow-up, in keeping with the lifestyle-related pathogenesis of disease.