Technical Abstract: Simulation models of soil related biological processes usually require soil temperature data. Frequently these soil temperatures are simulated and the soil temperature algorithms cannot be more complicated than the original process model. This situation has led to the use of semi-empirical type relationships in these process models. The objective of this study was to evaluate a hybrid soil temperature model, which combines empirical and mechanistic approaches, in an agroecosystem and a tallgrass prairie in the Great Plains. The original hybrid soil temperature model was developed and verified for a temperate forest system. This model simulated soil temperatures on a daily basis from meteorological inputs (maximum and minimum air temperatures) and soil and plant properties. This model was modified using different extinction coefficients for the plant canopy and ground litter. The agroecosystem consisted of a no-till corn (Zea mays L.) and soybean (Glycine max Merr. (L)) rotation system. Soil temperatures were measured at different depths in multiple years (three and two and a half years in the agroecosystem and tall grass prairie, respectively. In the agroecosystem, the root mean square error of the modified model simulation varied from 1.41 to 2.05 oC for the four depths (0.1, 0.2, 0.3 and 0.5 m). The mean absolute error varied from 1.06 to 1.53 oC. The root mean square error and mean absolute error of the modified model were about 0.1-0.3 oC less than the original model at the 0.2-0.5m depths. For the tallgrass prairie, the mean absolute errors of the simulated soil temperatures were slightly greater than the agroecosystem varying from 1.48-1.7 oC for all years and from 1.09-1.37 oC during the active growing seasons for all years.