Calculating evapotranspiration is still one of the main challenges. Most water and soil models, such as SWAT and WEAP, need to determine evapotranspiration for their atmospheric boundary conditions. This study evaluates the performance of 11 evapotranspiration (ET) models including three radiation-based methods, Makkink (Mak), Priestley-Taylor (PT) and FAO24 radiation (FAO24Rad), three temperature-based methods, Hargreaves Samani (HS), Blaney Criddle (BC) and Turc (Tu) and five combined models, Penman (P), ASCE Penman (ASCE Pen), Kimberly Penman (KP), FAO 24 Penman (FAO24Pen) and FAO 56 Penman-Monteith (FAO56PM) of differing complexity in a humid temperate environment in western Japan. The estimated daily ET were compared with measured evapotranspiration of Faba bean field at the first stage of growth, which was measured using a weighing lysimeter. Indicators, Nash–Sutcliffe efficiency (NSE), relative error (RE), and normalized root mean squared error (NRMSE) were used to evaluate the performance of potential ET estimations by the above-mentioned 11 methods. Bias indicators were also used for determining the overestimation or underestimation of methods. Results showed that all mentioned models underestimate the ET, NSE varies from -6.02 to 0.52, RE varies from 14 to 55%, and NRMSE varies from 0.016 to 0.063. The results of KP and FAO24Pen methods were close. The results of radiation-based methods were not satisfactory. Among the temperature-based methods, the HS method, with the 20% relative error, and from the five combination methods, FAO56PM, with the 14% relative error, showed better performances, respectively. Generally, the FAO56PM method has the highest accuracy respect to measured data.