Evidence-Based Medicine (EBM) have contributed to improved clinical practice with increased use of effective and life-saving treatments for severe diseases. However, the EBM model is less suitable for psychotherapy research than for pharmacological research and somatic medicine. The randomized controlled trial (RCT) design is an example of experimental methodology, which inevitably has more imperfections in psychotherapy research because psychotherapy RCTs cannot use double-blinding and the treatments tested are composite treatment packages. Long-term psychotherapy for severe and complex mental disorders is especially difficult to study with an RCT design. During the last decades, advanced analytic methods have been developed in psychotherapy process research, which enables investigation of causal connections regarding change mechanisms in psychotherapy. Therefore, we propose that the top of the research evidence hierarchy for psychotherapy should encompass: (1) RCT for circumscribed disorders, (2) cohort studies for complex disorders, and (3) advanced process studies for change mechanisms.