Background: Despite limited effectiveness of short-term psychotherapy for chronic depression, there is a lack of trials of long-term psychotherapy. Our study is the first to determine the effectiveness of controlled long-term psychodynamic and cognitive-behavioral (CBT) treatments and to assess the effects of preferential vs. randomized assessment. Methods/design: Patients are assigned to treatment according to their preference or randomized (if they have no clear preference). Up to 80 sessions of psychodynamic or psychoanalytically oriented treatments (PAT) or up to 60 sessions of CBT are offered during the first year in the study. After the first year, PAT can be continued according to the `naturalistic' usual method of treating such patients within the system of German health care (normally from 240 up to 300 sessions over two to three years). CBT therapists may extend their treatment up to 80 sessions, but focus mainly maintenance and relapse prevention. We plan to recruit a total of 240 patients (60 per arm). A total of 11 assessments are conducted throughout treatment and up to three years after initiation of treatment. The primary outcome measures are the Quick Inventory of Depressive Symptoms (QIDS, independent clinician rating) and the Beck Depression Inventory (BDI) after the first year. Discussion: We combine a naturalistic approach with randomized controlled trials(RCTs)to investigate how effectively chronic depression can be treated on an outpatient basis by the two forms of treatment reimbursed in the German healthcare system and we will determine the effects of treatment preference vs. randomization.