Solar energy is the most popular resource for power generation among the various available renewable energy alternatives. Solar radiation data are important for solar photovoltaic (PV) systems and passive energy-efficient building designs. Due to the unavailability of measurement in rural locations, solar radiation prediction models are required. In recent years, the Artificial Neural Networks (ANN) were successfully used for predicting solar radiation. However, previous works indicated that the ANN techniques were mainly focusing on prediction of monthly average or daily solar radiation, a few of them were modelled for predicting solar irradiance in hourly basis. In this study, prediction models of global solar irradiance on a horizontal surface will be developed based on neural-network techniques. Hourly meteorological variables between 2012 and 2015 acquired from the measurements made by local meteorological station were used for the study. To consider the effectiveness of individual predictors, different combinations of input variables were analysed using the Levenberg-Marquardt (LM) algorithm. Finally, equations were modelled by regression based on the important predictors. The developed models were used to estimate the global irradiance and assessed against measurement results.