The CCRT-method developed by Lester Luborsky is the most widespread and best established method to assess relationship structures in the field of psychodynamic psychotherapy research. Although its categorical structures are criticised in many different ways there have not been any attempts to modify them. This article points out the inconsistency of the current categorical system and demonstrates first approaches to alternative solutions. For this purpose genetic algorithms are used. Their application in psychotherapy research is demonstrated here for the first time.