Based on the actual sunshine hours, daily maximum and minimum temperatures data recorded at 98 meteorological stations from 1961 to 1999, the performance of six horizontal global solar radiation computational models with different meteorological factors is analyzed by simulating the observational materials, in which extraterrestrial solar radiation, dry and wet clean atmosphere radiation are included as starting data. The simulation result shows that the models with starting data of dry and wet clean atmosphere radiation can comparatively reflect the macro-terrain effect on the spatial distribution of solar radiation, while the model with starting data of only wet clean atmosphere radiation gives higher fitting precision. Among the 6 models, the analysis indicates that the two integrated models with the sunshine percentage as main variable, including the diurnal temperature range as correction variable, display the best performance; the model based on sunshine percentage has the best experiential coefficients stability and quite good precision, while the other three models based on the diurnal temperature range present the worst results. Finally, the model based on sunshine percentage is adopted to map the distribution of horizontal daily solar radiation in China, 2001.