The currently adopted approach for avalanche-hazard mapping in Switzerland includes avalanche-dynamics modelling coupled with estimation of the greatest annual 3 day snowfall depth, H₇₂, for high-return periods, used as the release depth. New advances in avalanche dynamics show that this approach can be improved using models with mass entrainment, requiring in turn a statistical definition of the erodible snow cover. We propose a regional approach, based on index value, to evaluate release depth and erodible snow cover for large-return periods. The territory of Switzerland is divided into seven climatologically homogeneous regions. Generalized extreme value (GEV) distributions for the growth factors coupled with index-value estimation based on altitude provide an accurate estimate of snow depths, also for large-return periods. RAMMS, a two-dimensional avalanchedynamics model including snow entrainment, is used for hazard mapping for a site used as an example of the Swiss procedure. The regional approach allows the boundary conditions for hazard mapping to be set using an entrainment model, and also provides statistical uncertainty of the design release and erosion depth, thus aiding in applying uncertainty analysis to hazard-mapping procedure.