AbstractCrop models can provide insights into the impacts of climate and management on crop growth and yield, but most currently are limited by overly simplistic assumptions about phenological development and response to water stress. We assessed winter wheat growth and yield performance of three crop models with lineage to the EPIC crop submodel. SWAT adopted the EPIC approach with few modifications, WEPS added new biomass accumulation, partitioning, and canopy approaches linked to key phenological development stages, and UPGM added to WEPS a detailed phenology component simulating responses to water-deficit stress. The models were evaluated with default parameters and compared to experimental data for winter wheat (Triticum aestivum L.) from two sites and a range of water-stress conditions for yield, aboveground biomass, biomass partitioning, canopy height, harvest index, and leaf area index. All models simulated yield very well (index of agreement [d] ≥ 0.93), but differences in model performance were increasingly evident for biomass (d = 0.91 [WEPS] to 0.86 [SWAT]), final canopy height (d = 0.68 [UPGM] to 0.44 [SWAT]), and harvest index (d = 0.61 [WEPS] to 0.43 [SWAT]). Errors in biomass simulation were most evident in the grain-filling period late in the growing season. Both WEPS and UPGM exhibited improved simulation of biomass and other response variables by including more explicit simulation of phenological response to water stress. The consistent improvement in winter wheat growth and yield simulation achieved with detailed phenology simulation provides an incentive to develop and test detailed phenology simulation components for other crops: currently 11 crops are simulated in UPGM, although the phenological parameters are uncalibrated. Better modeling linkages of water-stressed phenological development with other physiological processes will be critical to inform crop production where water stress and irrigation limitation are concerns.