Crop water deficits occur when transpiration demands from intercepted solar energy and from atmospheric water deficits cannot be met by water extracted from the root zone. The simulation of water deficits should therefore be based upon the processes whereby transpiration and extraction are determined, and whereby they interact to determine deficits, such that diurnal crop behavior is reproduced. However, the capabilities of this technique have not been widely tested. In order to simulate deficits based on these processes as part of an agroecosystem model, the calculated hydraulic resistances of the soil and roots were used to estimate the hourly rate of water extraction for a given canopy water potential. This rate was compared to that of transpiration under current meteorological conditions using canopy stomatal resistance calculated from the same canopy water potential. An iterative procedure was used to derive the canopy water potential for which uptake and transpiration were equal. The stomatal resistance associated with the canopy water potential was then used to estimate limitations to CO 2 uptake and growth as part of a sub-model of crop growth. The results of a maize irrigation study were used to test the response of the model, including the water deficit simulation, to different irrigation treatments at four different sites. Simulated and recorded values of evapotranspiration and dry matter yield ranged from 150 mm and 1 Mg ha −1 to 700 mm and 22 Mg ha −1. Simulated relationships between evapotranspiration and yield were consistent with those recorded, although simulated seasonal evapotranspiration at high irrigation rates was sometimes 5–10% higher than observed. Simulated seasonal water budgets relating yield, water use and soil water depletion to applied irrigation were also consistent with those recorded. This technique of simulating the dynamics of crop water deficits gives stable results over seasonal runs, and may be used to represent the diurnal behavior of crops under changing hydrologic conditions.