Abstract Estimates of precipitation from the Weather Research and Forecasting (WRF) Model and the Parameter-Elevation Regressions on Independent Slopes Model (PRISM) are widely used in complex terrain to obtain spatially distributed precipitation data. The authors evaluated both WRF (4/3 km) and PRISM’s (800-m annual climatology) ability to estimate frozen precipitation using the hydrologic model Structure for Unifying Multiple Modeling Alternatives (SUMMA) and a unique set of spatiotemporal snow depth and snow water equivalent (SWE) observations collected for the Olympic Mountain Experiment (OLYMPEX) ground validation campaign during water year 2016. When SUMMA was forced with WRF precipitation and used a calibrated, wet-bulb-temperature-based method for partitioning rain versus snow, its estimation of near-peak SWE was biased low by 21% on average. However, when SUMMA was allowed to partition WRF total precipitation into rain and snow based on output from WRF’s microphysical scheme (WRFMPP), simulations of snow depth and SWE were near equal to or better than simulations that used PRISM-derived precipitation with the calibrated partitioning method. Over all sites, WRFMPP and simulations that used PRISM-derived precipitation had relatively unbiased estimates of near-peak SWE, but both simulated absolute errors in near-peak SWE of 30%–60% at a few locations. Since, on average, WRFMPP had similar errors to PRISM, WRFMPP suggested a promising path forward in hydrology, as it was independent of gauge data and did not require SWE observations for calibration. Furthermore, in similar maritime environments, hydrologic modelers should pay close attention to decisions regarding rain-versus-snow partitioning, wind speed, and incoming longwave radiation.