After a long period of discussions on the efficacy of psychotherapy, there is still a lack of conventions for measuring change after psychological treatment. This paper first describes the concept of statistical and clinical significance of change. Using the SCL-90-R as a commonly administered instrument we then propose conventions and cut-off points for its global severity score (GSI) and change after therapy. A German standard population and several psychotherapy samples were aggregated to determine cut-off points and confidence intervals (reliable change indices) for statistically and clinically significant changes. Tingey et al. (1996) proposed the use of multiple clinical groups (inpatients and outpatients) aiming at a more realistic determination of "stepwise" changes. We examined this procedure with our data. Results show that it is not applicable in the German samples collected so far. Initial SCL-90-R scores in these groups did not differentiate sufficiently between inpatients and outpatients. Therefore, according to Jacobson and Truax (1991), moving from a "functional" to a "dysfunctional" population is still the criterion for a clinically significant change.