OBJECTIVE: (1) To use available research data to estimate the amount of change in eating behavior following obesity treatment; (2) To examine how this change relates to the amount of change in weight loss after treatment and at follow up. A meta-analysis was conducted in September 2009. METHODS: Studies were identified through a computer search of articles in the PubMed and PsychInfo databases. Key terms entered were obesity, treatment, and eating behavior. Effect sizes (Glass d) were calculated according to published procedures. RESULTS: Eighteen studies met the inclusion and exclusion criteria, grouped into two categories: psychosocial interventions and surgical interventions. For psychosocial interventions, we found a medium effect size on eating behavior (d=.73, CI=(.66, .90)), and a low effect size on weight (d=.32, CI=(.28, .36)) at posttreatment and a low effect size for both outcomes at follow-up (for eating behavior d=.47, CI=(.45, .49), for weight d=.37, (CI=.18, .56)). For surgical interventions we found large effect sizes on both outcomes (for eating behavior d=1.84, CI=(1.26, 2.42); for weight d=1.40, CI=(1.25, 1.65)). CONCLUSIONS: Surgical interventions have superior results to psychosocial interventions, on both weight loss and eating behavior. Implications for treatment of obesity are discussed.