Abstract. In this study two approaches are used to predict winter storm flow coefficients in meso-scale basins (10 km² to 1000 km²) with a view to regionalization. The winter storm flow coefficient corresponds to the ratio between direct discharge and rainfall. It is basin specific and supposed to give an integrated response to rainfall. The two approaches, which used the permeability of the substratum and dominating runoff generation processes as basin attributes are compared. The study area is the Rhineland Palatinate and the Grand Duchy of Luxembourg and the study focuses on the Nahe basin and its 16 sub-basins (Rhineland Palatinate). For the comparison, three statistical models were derived by means of regression analysis. The models used the winter storm flow coefficient as the dependent variable; the independent variables were the permeability of the substratum, preliminary derived dominating runoff generation processes and a combination of both. It is demonstrated that the permeability and the preliminary derived processes carry different layers of information. Cross-validation and statistical tests were used to determine and evaluate model differences. The cross-validation resulted in a best model performance for the model that used both parameters, followed by the model that used the dominant runoff generation processes. From the statistical tests it was concluded that the models come from different populations, carrying different information layers. Analysis of the residuals of the models indicated that the permeability and runoff generation processes did provide complementary information. Simple linear models appeared to perform well in describing the winter storm flow coefficient at the meso-scale when a combination of the permeability of the substratum and dominating runoff generation processes served as independent parameters.