The knowledge of the evapotranspiration of natural ecosystems and plant populations is of fundamental importance in several branches of science, research and practical uses. Nevertheless, the harmonisation of the large number of methods and user needs often causes problems. The objective of the analyses was to explore the output range and sensitivity of models of different physical approaches under local conditions. We performed descriptive statistical and sensitivity analysis of 10 commonly used estimation models - one of them with two variants. Correlation between modelled and measured evapotranspiration data series was assessed. The magnitude of the model outputs, their variability and responses to the changes of selected atmospheric parameters were evaluated. Priestley-Taylor, Penman-Monteith-FAO-56, Shuttleworth-Wallace (parameterized with alternative radiation balance), Szász and Makkink proved to be the most sensitive methods. As regards the systematic error, Makkink and Shuttleworth-Wallace showed the best agreement with pan evaporation, while Shuttleworth-Wallace, Blaney-Criddle and Makkink models were found to be the closest to the Penman-Monteith-FAO-56 method as a reference value.