A Richards-based soil water model was implemented in the APEX and EPIC terrestrial ecosystem models to improve their hydrologic modeling capabilities. The Richards model together with two existing soil water models were calibrated and evaluated to assess their performance for simulating watershed-level hydrology under scenarios of landscape conversion to bioenergy crop production. The Richards model was shown to better reflect observed soil-water dynamics in grain (corn) and cellulosic (switchgrass) bioenergy agroecosystems, whereas all three models simulated historic streamflows comparably. Application of the models to understand the impacts of widespread landscape conversion from traditional agriculture to bioenergy producing landscapes indicated disparate conclusions, with the Richards-based simulations indicating a modest 1.0% reduction in streamflow whereas the existing models simulated sizable reductions of 10.6–16.1%. This study clearly demonstrates the impact of model methodology on system understanding and contextualizes the wide range of simulated streamflow impacts from bioenergy conversions reported in the literature.