Hydrological models have become an important tool for the efficient management of water resources. However, selection of appropriate models for evapotranspiration (ET) computations in river basins remains challenging to watershed managers, especially in data-scarce regions. The performance of Soil and Water Assessment Tool (SWAT)-based model for the prediction of potential and actual evapotranspiration (PET and AET) of Ogun-Oshun river basin, Nigeria was investigated. Spatial and meteorological data was applied in setting up Map-window SWAT model. The three existing methods: Penman-Monteith, Priestly-Taylor (radiation-based) and Hargreaves (temperature-based), available in SWAT-were applied for the evaluation of PET and AET using soil, topographic, land-use and meteorological data as input parameters. The model results show a good correlation between the simulated and observed dataset as shown by Nasch-Sucliffe efficiency and coefficient of determination values. For the 30 year-simulation period, the predicted average PET values for Penman-Monteith, Priestley-Taylor and Hargreaves methods were 1791.516, 1684.597 and 1724.563 mm with corresponding standard deviation values of 89.322, 53.824 and 77.867 mm, respectively. The analysis indicated that Penman-Monteith and Hargreaves methods yielded almost the same results, while Priestly-Taylor method slightly differs, which establishes that it is not very suitable for arid/semi-arid regions. The study could be beneficial to watershed managers in addressing climate-related problems and for sustainable water resource management. © 2022 JUST. All Rights Reserved.