The original and enhanced cognitive model of eating disorders proposes that cognitive-behavioral therapy (CBT) "works" through modifying dietary restraint and dysfunctional attitudes towards shape and weight. However, evidence supporting the validity of this model is limited. This meta-analysis examined whether CBT can effectively modify these proposed maintaining mechanisms. Randomized controlled trials that compared CBT to control conditions or non-CBT interventions, and reported dietary restraint and shape and weight concern outcomes were searched. Twenty-nine trials were included. CBT was superior to control conditions in reducing shape (g=0.53) and weight (g=0.63) concerns, and dietary restraint (g=0.36). These effects occurred across all diagnoses and treatment formats. Improvements in shape and weight concerns and restraint were also greater in CBT than non-CBT interventions (g's=0.25, 0.24, 0.31, respectively) at post-treatment and follow-up. The magnitude of improvement in binge/purge symptoms was related to the magnitude of improvement in these maintaining mechanisms. Findings demonstrate that CBT has a specific effect in targeting the eating disorder maintaining mechanisms, and offers support to the underlying cognitive model. If changes in these variables during treatment are shown to be causal mechanisms, then these findings show that CBT, relative to non-CBT interventions, is better able to modify these mechanisms.