Abstract: Traditional arrangement methods of sampling points were established on the hypothesis that spatial characteristics of soil were random variation. But research results of geo-statistics showed that changes in spatial characteristics of soil were not completely random, soil characteristic status had relevance at spatial. So traditional arrangement methods were used for making sampling plan, those did not reflected local variation.. In this paper, classical statistical methods were used to determine the optimal number of sampling points, and geo-statistics methods, like semivariogram and Kriging methods, were used to determine the optimal arrangement of sampling points based on the analysis on spatial variation of soil fertility characteristics. Spatial variation of soil fertility was analyzed in typical areas. The results showed that 14 soil sampling points were enough to meet the needs of drawing prescription map for precision fertilization in the field of typical. It is a feasible method that classical statistics and geostatistical analysis were combined for optimal arrangement soil fertility sampling.