Evapotranspiration (Et) information is needed for many applications in agricultural and natural resource management, but commonly used prediction equations have not been adequately tested far the dry, high radiation, windy conditions of the Southern Great Plains. In order to test such equations, two precise weighing lysimeters at Bushland, TX, were used to measure daily Et from wellwatered, irrigated grain sorghum [Sorghum bicolor L.) Moench] in 1987 and 1988. Forms of the Penman, Penman‐Monteith, Jensen‐Haise, and Priestley‐Taylor models were evaluated for prediction of potential evapotranspiration (Etp). The Penman‐Monteith model provided the best predictions of Etp and predicted well across the entire range of measured Et. Forms of the Penman model with empirically fit wind functions overpredicted Etp by 20 to 409/0, and aftempts to fit a linear wind function to our data were not satisfactory The Jensen‐Hake model, requiring only daily mean temperature and solar radiation data, overpredicted by 30%, but performed consistently across the range of Et measured. Modification of the Priestley‐Taylor model with vapor pressure deficit or temperature advection terms is recommended. Without such modification, the model overpredicts at low Et rates and underpredicts at high Et rates.