Through an experiment of rasterization for 30 year mean air temperature and 1961's mean air temperature data from 617 meteorological stations in China with ARC/INFO and four interpolation methods, it was found that the Kriging method resulted in the highest precision, IDW method the second, Spline method the third, Trend method the lowest. The perennial mean air temperature and annual temperature anomaly can be used to rasterize annual air temperature data. This method will result in less analysis and computation but lower precision compared with the rasterization using annual mean air temperature data only. There is difference between meteorological stations' actual elevation and the DEM (Digital Elevation Model) values corresponding to the meteorological stations' longitudes and latitudes. Rasterization of air temperature data relies DEM fully. However, the model based on meteorological stations' actual elevation will result in higher precision for rasterization of air temperature data than that based on the DEM values corresponding to meteorological stations' longitudes and latitudes.