Potential evapotranspiration (PET) as the upper limit of the water vapor flux into the atmosphere is a fundamental and important concept in climate-related research. However, estimation of the PET using various approaches remains highly uncertain. Here we show the PET is overestimated by widely used Penman equation, as the underlying assumption of a saturated surface is not satisfied over most land surface. When the surface is unsaturated, the PET estimated from the aerodynamic approach (PETa) would exceed that from the energy balance approach (PETe), and the combination of the two in the Penman equation would lead to a systematic overestimation of the PET over land. This issue can be resolved with a modified Penman equation that accounts for the difference between PETe and PETa. Furthermore, we show that PETe as a function of the net radiation and a wet Bowen ratio is a good estimator of the PET based on the observationally constrained ERA5 reanalysis product. First, PETe can accurately estimate the ET over the ocean. Second, PETe is not sensitive to the surface moisture content, and the estimated PETe from observations under dry conditions approximates the PET well when the surface becomes saturated. Given the importance of the PET in meteorological, hydrological, ecological and agricultural research, the energy balance approach would have important implications for accurate estimation of the PET and its applications.