Abstract: Using the spatial variations of soil and the interpolation methods and with the sampling interval of 60m ×60m，the experiment gets 280 sample data about organic matter content，available P，available K ，total N ，total P and soon．And with the interpolation methods，such as inverse distance weighted，global polynomial，local polynomial，radial basic function，simple Kriging，universal Kriging and ordinary Kriging(exponential model，spherical model，Gauss model and stable mode1)，the soi1 available P of 140 samples are interpolated and the consecutive spatia1 distributions in different interpo lation methods are obtained．Comparing the prediction values with another 140 samples，the interpolation methods like local polynomial，global polynomial and three kinds of Kriging are better，and the exponential model，spherical model and Gaus model of ordinary Kriging are much better，and the local polynomial is the best one of all methods．