Abstract Daily values of insolation are important to many crop growth models. National Weather Stations that routinely record insolation are too sparse for adequate coverage of U. S. agricultural regions. Meteorological satellites can cover large areas with enough ground resolution and frequency to be used to estimate insolation. We developed a model using regression techniques on Geostationary Operational Environmental Satellite (GOES) and pyranometer data to predict insolation for Texas (winter 1976) and the Great Plains (summer 1977). Inputs required are GOES digital counts, solar zenith angle, and the visible clear radiance for each target. Correlation coefficients were 0.85 and 0.87 for the winter and summer data sets, respectively. Root-mean-square errors were 54.2 and 63.6 langley/day, respectively, within 11% of the summer mean insolation. Low insolation values caused by clouds were overestimated by the model. Comparisons with Tarpley's (1979) model are given.