<p>One of the most important challenges for the reference evapotranspiration (ET_o) estimation is the selection of the best model. Since all of the ET_o models are empirical, there is no an explicit benchmark model to evaluate the others. However, based on the pan evaporation for ET_o estimation, there is a principle that the pan coefficient (K_p) must be less than one. Regarding this principle as the benchmark, the aim of this study was to assess four famous ET_o models including FAO Penman–Monteith, Hargreaves–Samani, modified Hargreaves–Samani and Droogers–Allen based on the K_p indirectly derived from them according to the pan evaporation, and recalibrating the best of them for Iran’s northern and southern coastal regions. For this purpose, monthly data of air temperature, relative humidity, wind speed, solar radiation, precipitation, sunshine hours and pan evaporation were collected for a 30-year period (1994–2023). Results showed that the K_p derived from FAO Penman–Monteith, Hargreaves–Samani and modified Hargreaves–Samani was frequently more than one across most stations and months. In contrast, the Droogers–Allen model consistently yielded the K_p &lt; 1 and had the lowest error compared to the FAO-56 pan coefficients in both northern and southern coastal regions. Also, regional recalibration of this model based on the FAO-56 pan coefficients led to a considerable promotion (more than 100%) based on the Nash-Sutcliffe index in almost all stations.</p>