In the course of the upcoming new capital accord, Basel II, the different methodologies for the construction of internal rating models were in the center of interest within the last few years. Special attention was often put to the optimization of the discriminatory power and on statistical stability, frequently only obtainable by joining several portfolios. However, in practice it turned out that the discriminatory power itself is not a suitable measure for the quality of a rating system. Recently it became evident, that attention must be paid to the comprehensive validation of the systems just besides mature rating methodologies (Bundesbank 2003).