The 4-5 December event was simulated down to 1.33-km grid spacing using the both the Pennsylvania State University-NCAR Mesoscale Model (MM5 v3.7) and Weather Research and Forecasting (WRF v2.1) model using the NCEP GFS for initial and boundary conditions, Mellor-Yamada-Janjic (Eta) PBL, Kain-Fritsch convective parameterization on the 36- and 12-km grids, and the Thompson microphysical parameterization, which includes graupel processes. Both MM5 and WRF realistically simulated the three dimensional thermodynamic and kinematic fields at 1.33-km grid spacing, including 1-2 m/s vertical velocities produced by the narrow windward ridges. However, both models overpredicted surface precipitation over portions of the lower windward slope and immediate lee of the Cascade crest, which is similar to other IMPROVE-2 case studies. Microphysical comparison with NOAA P-3 in situ measurements showed that both models produced 2-3 times as much snow as observed at 2-3 km ASL over the Cascades using the Thompson bulk microphysical scheme, while cloud water was overpredicted above some of the windward peaks. The various bulk schemes within MM5 and WRF were also tested, with sensitivity simulations used to explain some of the fundamental bulk scheme differences.