A quantitative analysis of evapotranspiration was carried out at the University of North Dakota Meteorological Station. From data collected on a daily basis during the summer of 1967, several linear regression models predicting evapotranspiration were derived. The variables used in the study were actual evapotranspiration (dependent), air temperature, relative humidity, air movement, water loss from an open pan, and soil temperature (independent). The model providing the best fit yielded a multiple correlation coefficient of R = .49321. A simplified predictive equation for evapotranspiration was obtained, using only pan loss and soil temperature as independent variables. In this case, R = .45792, indicating little loss in prediction. The single variable correlating most highly with evapotranspiration was maximum air temperature, for which R = .4291. These values are interpreted as surrogates for actual evapotranspiration information. Better prediction was achieved in the case of estimating water loss from an open pan. Using relative humidity, total air movement, maximum air temperature, and soil temperature as independent variables, a multiple correlation, R = .78954, was obtained. (Author)