In cold climate regions a significant fraction of annual soil erosion in agricultural land occurs during snowmelt and rain on partially frozen soils. Physically based and spatially distributed soil erosion models have proved to be good tools for understanding the processes occurring at catchment scale during rainfall erosion. However, most existing erosion models do not account for snow in a suitable way. A combination of the UEBGrid snow pack model and the LISEM erosion model was therefore used in this study. The aim was to test and validate this model combination and to assess its utility in relation to quantification and process understanding. Applying this model combination to simulate surface runoff and soil erosion showed that, in principle, it is possible to satisfactorily simulate surface runoff and observed soil erosion patterns during winter. The values for the calibration parameters were similar for the two chosen winter periods when the rainfall and snowmelt episodes occurred. However, the calibration procedure showed that the utility of this combination had several limitations. It is hoped that this study can help to improve existing models and trigger new developments in including snow pack dynamics and soil freezing and thawing in soil erosion models.