Abstract: The research took Tongling County as the study area, apply Geostatistics combined with GIS, selected the soil organic matter as research object, and analyzed the spatial variability feature. According to Kriging interpolation theory, the study further applies different semivariogram model and compares the accuracy between the Root Mean Square Standardized Error (RMSSE) and the Agreement Coefficient (AC). The research indicated that, the spatial interpolation by log transformation cannot improve accuracy, using the RMSSE to compare the accuracy with the original interpolation sample was only a reference, independent validation samples were necessary. The coefficient of variation of soil organic matter was general, the content distribution was uneven and the higher organic matter distributed in the high-altitude location, influenced mainly by the terrain of the region.