Hydrologic models provide the land-phase link between atmospheric models and oceanographic models within the global water cycle. They provide independent validation of the outputs from atmospheric models and may also provide a mechanism to examine the implications of climatic change on water resources. To form a component of a global model, hydrologic models must be applicable at macroscale and continental scale. The sources and scales of input data are critical to the development of such models. Results from the application of a distributed hydrologic model to Canadian watersheds from 500 km2 to 1.6 million km2 in area are used to compare errors found using input data at different scales. Data considered include land cover, vegetation index, and snow water equivalent from satellite sensors and distributed climate data from a general circulation model and from numerical weather prediction models. Analysis of the results allows a consideration of appropriate data scaling in the development of macroscale hydrologic models. ¿ American Geophysical Union 1995