The estimation of the spatial distribution of mean monthly soil temperature in January,April,July and October at a depth of 5cm in Beijing,Tianjin and Hebei district was made by two approches. One is with a regression model in which the predictors of Latitude,Longitude and Sea Level are involved. The other is with the Kriging surface fitting method. The results show that the standard deviation of residual errors of the regression model is 0. 4-0. 7℃,the standard deviation of residual erorrs of the Kriging estimation is 0. 2-0.3 ℃. In addition,as the value of latitude incrises by one degree soil temperature decrises by 0. 2-0.8℃,as the value of longitude incrises by one degree soil temperature decrises by 0. 3-Al-0.7℃, as the value of sea level incrises 100 meters the soil temperature decrises by 0. 5-0. 7 ℃.