Accurate estimation of spatial distributions of evapotranspiration (ET) is a goal sought by hydrologists, agronomists, and meteorologists but is difficult to achieve. The usual approaches to estimating ET employ remote sensing observations and a surface energy flux model. However, resolution of remote sensing data, needed to observe patterns of biophysical variables, is commonly too coarse (>1 km) to distinguish between land cover types that constrain ET. Accuracy of ET estimates can be improved by using higher-resolution (<100 m) remote sensing data since they can distinguish clusters of vegetation from bare soil fields and water bodies. A demonstration of this potential is shown using aircraft-based remote sensing observations over a study site at El Reno, Oklahoma. Five midday surveys, conducted from 29 June to 2 July 1997, as part of the Southern Great Plains 1997 Experiment (SGP97), collected 12 m resolution images in the visible, near infrared, and thermal infrared. Surface temperature and vegetation density maps, created from these surveys, were combined with surface micrometeorological observations and with a two source energy balance model. Results from El Reno show that flux estimates with respect to ground-based eddy covariance observations can be accurate to within 40--80 W m-2. This means that the high spatial resolution observations can potentially produce ET estimates similar in quality to ground-based point measurements. Additional work, needed to show how high-resolution remote sensing estimates can be related to coarser resolution observations, is underway using the satellite sensors ASTER (15--90 m resolution) and MODIS (250 m to 1 km resolution).