An expedient and accurate method for assessing the effect of intensive agriculture on a watershed water balance is needed, particularly for those areas where well irrigation is prevalent or becoming more so. Digital image processing of Landsat-3 data discriminated major field crops and forests in Tift County, Georgia, necessary for a water use assessment. A water balance model based on a modified Penman relationship was developed to estimate evapotranspiration. The remotely sensed data and local meteorological data were then used in a hydrologic model to predict watershed daily water balance. The simulation was evaluated by comparing model performance with water balance measurements for a 15.5 km2instrumented agricultural watershed.