Within a naturalistic design 44 psychoanalytic treated patients are examined with regard to qualitative and quantitative outcome. The results are compared to 56 dynamic and 164 inpatient therapies. A comparison of symptoms, diagnoses and motivation before therapy leads to the conclusion that there are very different patient groups treated within this different settings. Therefore randomization seems to be an inadequate strategy to compare groups in different therapeutic settings. Using different criteria of outcome it can be demonstrated that psychoanalytic treated patients improve very well and to a larger extent than psychodynamically treated patients or inpatients. Different criteria of outcome and different observing perspectives (patient/therapist) lead to interesting results. Patients report primarily improvements in somatic, anxiety and depressive complaints. Therapists moreover report substantial improvements in interactional symptoms and behaviour. The different results in using several criteria are presented and discussed with regard to the methodology of outcome measurement. Finally recommendations for further studies in the psychoanalytic context are presented.