The PILPS 2(d) experiment at Valdai offers a unique opportunity to evaluate land surface schemes in a climate with seasonal snow cover and seasonally frozen soil. Here we evaluate the 18-year simulations by the 21 schemes that participated in the experiment, in their ability to correctly partition the melting snow in the spring into runoff and infiltration. Using observed vertical profiles of soil temperature and soil moisture, we evaluate the relationship between infiltration and the frozen layer in the soil, and determine the dependence of this modeled relationship on the correct partitioning of snow meltwater. The large interannual variations allow us to provide new insights into the level of complexity necessary to correctly model this important aspect of land surface hydrology