Cognitive-behavioral therapy (CBT) is the best examined treatment for eating disorders. However, previous meta-analyses of cognitive-behavioral therapy have not examined absolute outcomes, which are important from a clinical perspective. We updated a meta-analysis and conducted new searches in PubMed, Embase, PsycINFO and CINAHL. We included randomized trials comparing CBT with control conditions in adults with a diagnosed eating disorder. We used random effects models in all analyses. We included 36 trials with 44 comparisons between CBT and controls (2,809 participants), 22 trials on binge eating disorder (BED), 11 on bulimia nervosa (BN), and three on anorexia nervosa and mixed disorders. The overall effect of CBT compared to controls was g = 0.88 (95% CI: 0.71; 1.04), with high heterogeneity (I2 = 74; 95% CI: 65; 81; PI: -0.06; 1.81) and no significant difference between BED and BN. Effects were smaller studies with low risk of bias. The absolute abstinence rate was 0.36 (95% CI: 0.31; 0.43) for CBT and 0.10 (95% CI: 0.08; 0.12) in controls. CBT is probably effective in the treatment of bulimia nervosa and binge-eating disorder, but there is also a large group of patients who do not respond sufficiently.