Land managers need to include climate change in their decisionmaking, but the climate models that project future climates operate at spatial scales that are too coarse to be of direct use. To create a dataset more useful to managers, soil and historical climate were assembled for the United States and Canada at a 5-arcminute grid resolution. Nine CMIP3 future climate projections were downscaled to this grid and the MC1 dynamic global vegetation model was run over the historical and future climates. Climate variables included monthly mean temperature, precipitation, minimum temperature, maximum temperature, and vapor pressure. Soil data included soil depth as well as percentages of sand, clay, and rockiness for three soil depths. Output variables included carbon pools and fluxes, fire variables, potential vegetation classes, and various water cycle variables. Climate and soil inputs as well as MC1 outputs are available publicly.