Practitioners and researchers alike assume that there is individual variability in the effects of treatments for mental disorders. However, for psychotherapy, up to now this assumption has never been empirically tested. Using a large database of randomized-controlled trials on psychotherapy of depression in adults (306 trials including a total of 51,853 patients), we performed a Bayesian variance ratio metaregression. For the entire sample, we found a 9% higher variance in the intervention groups compared with the control groups. Depending on the depression scale used, this corresponds to a standard deviation of the individual treatment effect of 3 to 4 points. Subgroup analyses revealed that the effect variability of some types of therapy is larger than others. Our results are the first to indicate that patients do benefit differently from psychotherapy. We conclude that there is a sound basis for the paradigm of personalized psychotherapy, which brings about implications for both research and clinical practice. Public Health Significance StatementIn recent years, studies with high methodological quality have pointed out that the efficacy of psychotherapy in the treatment of depression is less satisfactory than previous research suggested. To optimize psychotherapy for non-responders, the paradigm of personalized therapy is coming into the research focus. In this study, we show for the first time that the effects of psychological interventions vary more than those of control conditions. This shows that differential response to treatments is inherent to intervention effects. Thus, it could indeed be beneficial to better tailor psychotherapy to individual patients. In practice, session-by-session outcome monitoring should be used to detect non-responding cases in ongoing treatments. Statistical methods guiding the selection of treatment components capitalize on the heterogeneity of treatment effects and are thus likely to improve outcomes. These findings pave the way for broad research and implementation of approaches that support personalization (e.g., monitoring and feedback systems) as well as a new shaping of training beyond the traditional schools of thought in psychotherapy.