This study evaluated the potential of using the Surface Energy Balance Algorithm for Land (SEBAL) as a means for estimating evapotranspiration (ET) for local and regional scales in Southern Idaho. The original SEBAL model was refined during this study to provide better estimation of ET in agricultural areas and to make more reliable estimates of ET from other surfaces as well, including mountainous terrain. The modified version of SEBAL used in this study, termed as SEBALID (lD stands for Idaho) includes standardization of the two SEBAL "anchor" pixels, the use of a water balance model to track top soil moisture, adaptation of components of SEBAL for better prediction of the surface energy balance in mountains and sloping terrain, and use of the ratio between actual ET and alfalfa reference evapotranspiration (ETr) as a means for obtaining the temporal integration of instantaneous ET to daily and seasonal values. Validation of the SEBALID model at a local scale was performed by comparing lysimeter ET measurements from the USDA-ARS facility at Kimberly, Idaho, with ET predictions by SEBAL using Landsat 5 TM imagery. Comparison of measured and predicted ET values was challenging due to the resolution of the Landsat thermal band (120m x 120 m) and the relatively small size of the lysimeter fields. In the cases where thermal information was adequate, SEBALID predictions were close to the measured values of ET in the lysimeters. Application of SEBALID at a regional scale was performed using Landsat 7 ETM+ and Landsat 5 TM imagery for the Eastern Snake Plain Aquifer (ESP A) region in Idaho during 2000. The results indicated that SEBALID performed well for predicting daily and seasonal ET for agricultural areas. Some unreasonable results were obtained for desert and basalt areas, due to uncertainties of the prediction of surface parameters. In mountains, even though validation of results was not possible, the values of ET obtained reflected the progress produced by the refinements made to the original SEBAL algorithm.