Studies using data from a general circulation model (GCM) and from numerical weather prediction models (NWP) as inputs to a hydrological model have produced useful results but were limited by the coarse resolution of the GCM data on the one hand and, on the other hand, by the relatively short period of high-resolution NWP data available. This paper describes an alternative approach using long-term (96 years) daily data from a high resolution atmospheric boundary layer model. The data from the atmospheric model were used as inputs to a daily distributed hydrological model of the Upper Columbia Basin in western Canada. The hydrological model was calibrated for five years and verified using streamflow and other data for a further 18 years. It was found that the hydrological model was able to simulate snowpack and streamflow data for the total period of 96 years of available atmospheric model output with good success. This method provides the opportunity for analysis of water resources systems under alternative climate and land-use scenarios.