Prediction of solar radiation is needed for a wide range of applications, such as forpredicting daily evapotranspiration. In our case, prediction of solar radiation is important forestimating power and energy needs. Predicting clear day radiation is a clear process, while predicting cloudy day radiation can be more complex. Having a capability to predict solar radiation forlocations with no or very few data is an important need that we are trying to address. TheHargreaves and Samani model estimates for daily solar radiation in Wooster OH had goodcorrelation between model prediction error and average solar radiation throughout year. To conducta more appropriate assessment of results, we found that the RMSE should be calculated for a dry, ahumid, and an average year, instead of calculating a general RMSE for the whole period. Choosingthe model constant based on the difference between max and min daily temperature can improvepredictions. Several modified versions of the Hargreaves and Samani model, such as the Allen selfcalibratingmodel, a lower-limit calibrating model based on relative humidity, and severalcombinations of these models, were used to evaluate using the models to produce probabilitydistributions for solar data, and predicting statistical upper and lower limits of data in Ohio. The erroranalysis shows that the models with a combination of self-calibrating for the upper and lower solarradiation were more successful in predicting the expected solar limits.