Accurate prediction of such extreme snowstorms is essential to protect the safety and well-being of the public and mitigate infrastructural and socioeconomic damages, specifically power outages. Numerical Weather Prediction (NWP) of snowstorms faces challenges related to microphysical, initial and boundary layer processes that hinder the accuracy of forecasts. In addition, observations of mixed phase precipitation are very sparse spatially and temporally which causes an added challenge to validate predictions. We present an evaluation of extreme snowfall prediction for 34 significant NE US snowstorms from 2006 to 2017 using three NWP models (RAMS/ICLAMS, WRF 3.7, and WRF 3.8). Snow storm prediction is evaluated using 109 Meteorological Terminal Aviation Routine Weather Reports (METAR) and available radiosonde stations. Snow events have been identified due to their impacts on the electrical power grid and the variables considered for this evaluation are associated with winter storm severity, precipitation type, snowfall ratios, and liquid to ice ratios.