Predicting Canadian mountain headwaters hydrology is highly dependent on accurately simulating the dynamics of the seasonal snowcover. Mountain MESH, a Cryospheric-Hydrological Land Surface model built around a Canadian Land Surface Scheme core, represents blowing and intercepted snow redistribution, sublimation, and snowmelt as influenced by elevation, slope and aspect, infiltration to frozen soils, and the full suite of land surface hydrological processes. Mountain MESH’s ability to predict streamflow, snow water equivalent (SWE) and snow-covered area (SCA) was evaluated in a small instrumented subalpine headwaters basin in the Canadian Rockies. The model was forced using observations from a network of meteorological stations in the basin and parameterized using basin attributes, without calibration. Streamflow predictions were evaluated from hydrometric observations over the spring freshet through autumn. SWE simulations were evaluated in needleleaf forests, forest clearings and alpine tundra using snow depth and density survey transects, and across the basin using LiDAR-derived snow depths. SCA simulations across the basin were evaluated using Sentinel satellite SCA.  The results showed that model performance in predicting streamflow was excellent, indicating good water balance closure.  SWE predictions were better in wind-sheltered than in wind-exposed landcovers. SCA predictions were best in the evergreen needleleaf forests and shrubs and less accurate in the deciduous needleleaf forests. Model falsification by suppressing blowing snow redistribution caused substantial deterioration in predictive performance, demonstrating the need to include both snow redistribution and ablation physics in mountain snow hydrology models.