The U.S. Army Research Laboratory has developed a mesoscale model; the Battlescale Forecast Model (BFM), which is a hydrostatic model designed for boundary-layer applications for the Army. The BFM is a 16-layer terrain-following model with a top of 7000 m above the highest elevation on the grid. The BFM is a single-nest model with a horizontal resolution of 10 km. While the model does not contain moist physics or cumulus parameterization packages, the non-convective precipitation rate is parameterized as a function of cloud liquid water. Using the scheme developed by Sundqvist in 1989, the basic assumption in this statistical cloud model is: the denser the cloud the higher the precipitation rate. Since ARL receives 15-km MM5 output from the Air Force Weather Agency, it was decided to study the skill in forecasting stratiform precipitation, precipitation rates, biases, and precipitation types from each of the two models. The precipitation type forecast is a post-processed routine developed by ARL and is used on both the BFM and MM5. The key parameters to determine if precipitation will fall as rain, snow. freezing rain, or mixed precipitation are the wet bulb temperature, depth of the freezing layer, temperature advection, boundary-layer temperature, surface temperature, and radiation amounts that penetrate through a cloud layer. Results of a study during the winter months of 2003 indicate that both models forecast much weaker precipitation rates for snow than rain. Both models showed excellent results in determining the precipitation types rain and snow, although they had a more difficult time with the complexity of freezing rain and mixed precipitation. Apparently, based on this study, the model biases and model numerics do play a significant role in determining the precipitation rates, intensity, and precipitation types.