A cloud-masked fractional snow-covered area (SCA) product gridded at 1 km was developed from the advanced very high resolution radiometer for the Colorado River and upper Rio Grande basins for 1995-2002. Cloud cover limited SCA retrievals on any given 1-km(2) pixel to on average once per week. There were sufficient cloud-free scenes to map SCA over at least part of the basins up to 21 days per month, with 3 months having only two scenes sufficiently cloud free to process. In the upper Colorado and upper Grande, SCA peaked in February-March. Maxima were 1-2 months earlier in the lower Colorado. Averaged over a month, as much as 32% of the upper Colorado and 5.5% of the lower Colorado were snow covered. Snow cover persisted longest at higher elevations for both wet and dry years. Interannual variability in snow cover persistence reflected wet-dry year differences. Compared with an operational (binary) SCA product produced by the National Operational Hydrologic Remote Sensing Center, the current products classify a lower fraction of pixels as having detectable snow and being cloud covered (5.5% for SCA and 6% for cloud), with greatest differences in January and June in complex, forested terrain. This satellite-derived subpixel determination of snow cover provides the potential for enhanced hydrologic forecast abilities in areas of complex, snow-dominated terrain. As an example, we merged the SCA product with interpolated ground-based snow water equivalent (SWE) to develop a SWE time series. This interpolated, masked SWE peaked in April, after SCA peaked and after some of the lower-elevation snow cover had melted.