Two land surface process models (LSM version 0, LSX version 1.02) with different parameterizations of energy exchange, turbulent transfer, and hydrology were compared to examine the climatic importance of these differences. Simulations were for three surface types (tropical forest, temperate grassland, tundra) using prescribed atmospheric forcings. Although the two models used quite different canopy energy exchange parameterizations, especially turbulent transfer, most differences between the two models were related to soil hydrology. LSM generally had greater infiltration and less surface runoff than LSX. As a result of the different fluxes of water into the soil, the models differed greatly in sensible and latent heat fluxes. These differences in infiltration, surface runoff, latent heat flux, and sensible heat flux were directly related to the parameterization of infiltration capacity. The two models differed in stomatal resistance, and when given the same stomatal resistances, both models had similar transpiration rates. Soil evaporation differed between models, apparently due to differences in the temperature calculations. Snow albedo also differed between models. These comparisons suggest that future development of land surface models should pay more attention to soil and snow hydrologic processes.