Abstract: In this paper,the prediction model of surface solar radiation is established based on the multiple stepwise regression method. First,with the data from 1993 to 2013 in the database of the China Meteorological Science Data Sharing Service,the correlation between the amount of solar radiation and the duration of sunshine is analyzed. And then,with multiple stepwise regression analysis,the multiple regression models used to forecast the solar radiation exposure are obtained. Furthermore,a univariate regression model is established to extract the scattered radiation from the total radiation. Finally,the calculation results of the model are compared with the actual data to validate the effectiveness and accuracy of the proposed model,which can provide reference for the research of planning and design,modeling and simulation and power prediction of photovoltaic（PV） power stations.