680 unselected outpatients of our Psychosomatic Clinic were divided into six diagnostic groups and examined by the SCL-90-R. By ANOVA we found differences of means between the diagnostic groups for the global severity score (GSI) and for the subscale scores. We describe the model of Jacobson et al. [1,2] for the determination of cut-off-points and reliable change indices for the assessment of changes over time, e.g. after psychotherapy. Our results lead to the conclusion that, although an overall reliable change index for the GSI can be applied to all psychosomatic patients, cut-off-points should be different following the diagnostic groups. Furthermore, information from the SCL-90-R-subscales as well as other specific symptom scales should be used for the interpretation of significant changes of the GSI.