There is no mental disorder for which the effects of psychological treatments have been examined so elaborately as depressive disorder. There are at least 200 controlled and comparative efficacy studies. Most of this research focuses on cognitive and behavioral therapies but there also many studies examining the effects of interpersonal psychotherapy, problem solving therapy, social skills training, and psychodynamic therapies. These therapies have been examined for adults with depressive disorders and high levels of depressive symptoms, but also for several specific populations such as children, adolescents, elderly, women with postnatal depression, and people with various somatic illnesses, such as stroke, heart disease, and diabetes. Furthermore, the therapies have been delivered in several formats, such as group treatment, individual treatment, bibliotherapy, and internet-therapy. A meta-analysis is the best method to integrate these results. In the past ten years, about 22 of those meta-analyses examining the effects of psychological treatment of depression have been conducted. All of these meta-analyses, however, focus only on a specific method or target population. Modern meta-regression methods, in which moderators and mediators of outcome can be analysed, has not yet been conducted. In the present project, we propose to conduct a new, and comprehensive meta-analysis of psychological treatment of depression. Such a project is important to establish which type of therapy is most effective for whom. One probable reason why no comprehensive meta-analysis has been conducted until now, is that the target groups, the psychological interventions, and the formats in which the interventions are delivered, differ very much. In a meta-analysis, this results in heterogeinity, indicating that the included studies differ systematically. And because of these systematic differences, direct comparison in a meta-analysis is problematic. However, we are planning to conduct a meta-regression analysis, in which these differences are the focus of the research. We assume that all psychological treatments of depression are effective, and if we find heterogeinity, we examine the sources of heterogeinity. These sources are exactly the mediators and moderators we want to discover. Comparable studies in other research areas have shown that this approach is very useful, and results in important new insights in intervention research. Another reason why such a metaregression study has not been conducted until now, is that most meta-analyses have included relatively small number of primary studies (the largest had included 63 studies. These small numbers of studies result in insufficient statistical power for conducting metaregression analyses. In the proposed project, however, we expect to include at least 200 controlled and comparative studies, which is sufficient for examining mediating and moderating variables in a metaregression study. Two researchers will conduct systematic searches in several databases with selected keywords to identify primary studies. Furthermore, the references of the included studies will be examined. We will include studies (-) in which a psychological treatment for (-) subjects with depressive disorder or a high level of depressive symptomatology (-) is compared to a control group and/or another treatment. From each study characteristics will be collected with regard to: population, intervention, design and analyses, and measures. Lastly, the methodological quality of the included studies will be assessed with the quality rating scale developed by Moncrieff et al. This will be done independently by two reviewers.