It is well established that potential evapotranspiration (PET) can be reliably estimated using the energy budget at the canopy or land surface. However, in most cases the necessary measurements are not available. Because of this, many mostly-empirical temperature-based models have been developed and are widely used. Here we test whether a radiation based model (Priestley-Taylor) can reliably predict PET using air temperature to estimate the radiation fluxes. We used data from the AmeriFlux network to approximate net radiation from daily minimum and maximum temperature measurements, day of the year, and geographic location of the sites; i.e., readily available data in most places. We found good agreement between Priestley-Taylor PET calculated from measured radiation fluxes and Priestley-Taylor PET determined primarily via air temperature. The most difficult parameter to estimate was the atmospheric transmissivity to in-coming solar radiation. Overall the results suggest that radiation-based PET esti-mates can be made even when direct measurements of the radiation fluxes are unavailable.