Mountain snowpack spatial variability can impact snow-driven streamflow timing and magnitude. Near-peak, lidar-derived snow water equivalent (SWE) data were coarsened (50 m to 1 km; elevation bands), re-gridded, and inserted into the Distributed Hydrology Soil Vegetation Model (DHSVM) to evaluate how snow heterogeneity at peak SWE influences streamflow in the Tuolumne River, California, and Blue River, Colorado. In the Tuolumne, peak simulated streamflow from the most homogeneous snowpack representation was 36% greater than heterogeneous simulations. In all years, the homogeneous snowpack representation did not produce a recession limb peak since snow melted more quickly. Streamflow during late summer was higher (up to 3.5 times) for finer-resolution snow heterogeneity simulations. In the Blue, streamflow simulations were more similar across different snowpack representations, and the sensitivity to heterogeneity was less pronounced. This difference between watersheds was attributed to differences in SWE variability and subsurface processes between the two basins. In the Blue, snowmelt infiltrated into the subsurface and was slowly released throughout summer. In the Tuolumne, mid-winter melt and rain resulted in higher soil moisture at peak SWE. The combination of high soil moisture and the Tuolumne’s subsurface characteristics allowed water to move more efficiently through the subsurface than in the Blue. For both watersheds, altering snow heterogeneity had little effect (within 2%) on total end-of-water-year runoff or evapotranspiration because absolute differences balanced out throughout the melt season: a higher-resolution SWE representation produced less streamflow at peak SWE but greater streamflow afterward relative to a coarser SWE representation. Experiments isolating radiation variability demonstrated that fine-scale accumulation processes at peak SWE had a stronger influence on streamflow timing than post-peak radiation heterogeneity. These results suggest that if basin-mean SWE is well constrained, capturing heterogeneity may not be necessary for water supply forecasting on quasi-annual timescales but can be critical for simulating late summer flows.