The findings of a joint ISSMGE TC 205/TC 304 discussion group on the evaluation of model uncertainties, i.e. uncertainties related to the chosen geotechnical design method, are presented. The discussion revealed the importance of model uncertainty assessment due to the high sensitivity of many geotechnical design methods for the accuracy of the adopted model. In case of geotechnical designs in which the design quantities are clearly defined, model uncertainties may be considered by a model factor representing the bias of the predicted to the real behavior. The real behavior itself can be derived from well documented prototype and model scale field and laboratory tests. Model factors can be used in rigorous reliability based design but also in LRFD formats adopted by many geotechnical design codes. However, for complex design situations model uncertainty assessment remains difficult. Moreover, model uncertainties are affected by extraneous uncertainties such as spatial soil variability or uncertainties due to test performance which cannot be separated.