Based on daily air temperature data from 27 meteorological stations in Shanxi plateau and its surrounding areas, in combination with digital elevation model,the annual average temperature of Shanxi plateau from 1981 to 2005 was interpolated using spatial interpolation methods including inverse distance weight,global polynomial interpolation,local polynomial interpolation,radial basis function,spline,trend surface,ordinary Kriging,and Co-Kriging.The precision of these methods was evaluated through leave-one-out cross-validation.The spatial distribution of temperature in Shanxi plateau was explored by the optimal method.The results showed that: 1) DEM-adjusted Co-Kriging method is optimal for interpolating the temperature values of Shanxi plateau; 2) The accuracy of Co-Kriging and ordinary Kriging based on geostatistics was higher than that of general interpolation methods; 3) The accuracy of general interpolation methods in descending order was radial basis function method,local polynomial method,inverse distance weighting method,global polynomial method,trend surface method and spline function method; 4) The spatial distribution of air temperature in Shanxi plateau presented regional characters,which showed a decrease trend from low latitudes to high latitudes.