Psychotherapeutic research is generally carried out to investigate the effectiveness of specific techniques with selected, often homogeneous samples of patients. This leads to problems when applying research findings to the clinical reality where the patient population is heterogeneous. In this study, we examine psychotherapy outcome in a natural treatment setting using criteria established by Jacobson and his colleagues for measuring improvement in psychotherapy, reliable change and clinical significance. Outcome was evaluated in a sample of 117 Dutch ambulatory mental health center patients with the SCL-90 Symptom Checklist and the Target Complaints List. On both instruments improvement rates varied greatly and were only moderately correlated. Our findings suggest that a more limited application of the Jacobson criteria derived from selected samples is warranted when applied in a general patient population.