A well-known approach to seasonal hydrological forecasting involves the use of ensembles as forcing to a hydrological model, based either on historical meteorological data (Ensemble Streamflow Prediction, ESP) or on forecasts produced by one or more dynamic climate models (multi-model approach). This work aims at investigating the role of initial hydrologic conditions (ICs) and seasonal climate forecast skill on the accuracy of seasonal hydrologic predictions in alpine regions, as a function of catchment properties. The Integrated Catchment-scale HYdrological Model (ICHYMOD) is employed, forcing it with historical meteorological data and multi-model ensemble climate predictions produced by the climate forecast systems NCEP CFSv2 and ECMWF SEAS5 (the latter analysis is not included in the present work). The hydro-climatic prediction system is tested on two catchments in the Eastern Italian Alps, that are different in terms of orography and soil/groundwater storage capacity. The diverse catchment properties result in differential parameterization of the subsurface processes in the hydrological model, hence in a different impact of the initial hydrologic conditions on the seasonal runoff predictions.