The snowmelt runoff model (SRM) uses a degree-day approach for melting snow in a basin. A simple radiation component was combined with the degree-day approach (restricted degree-day method) in an effort to improve estimates of snowmelt and reduce the need to adjust the melt factor over the ablation season. A daily energy balance model was formulated that requires not only the input of radiation but also measurements of daily wind speed, air temperature, and relative humidity. The three approaches for computing snowmelt, namely, the degree-day, restricted degree-day, and daily energy balance model were tested at the local scale by comparing melt rates with lysimeter outflow measurements. Because radiation measurements are not often available, a simple model for simulating shortwave and longwave components of the radiation balance that requires minimal information (i.e., daily cloud cover estimates, air temperature, and relative humidity) was developed. It was found that clouds and their effects on daily isolation at the surface can produce significant differences between measured and model estimates. In the comparisons of snowmelt estimates with the lysimeter outflow, the restricted degree-day method yielded melt rates that were in better agreement with the observed outflow than the degree-day method and were practically the same as estimates given by the energy balance model. A sensitivity analysis of runoff generated with SRM using as input the local snowmelt computations given by the three models and measured outflow from the lysimeter was performed for a basin. A comparison of the synthetic hydrographs for the basin suggests that a radiation-based snowmelt factor may improve runoff predictions at the basin scale. ¿ American Geophysical Union 1994