In an agricultural ecosystem, soil organic matter (SOM) is an index key to estimating soil C sequestration, soil fertility and quality, etc. Estimation of these soil properties at an acceptable level of accuracy is very important. In this research, a square area (116^°45′32″~117^°2′14″E，34^°31′12″~34^°45′6″N), about 580 km2, was chosen as an example in the northwest of the Xu-Huai alluvial plain (Xuzhou and Huaiyin). A total of 168 soil samples were collected according to nested scenario’s, for analysis of SOM content and soil texture. And furthermore, spatial variability of SOM in the surface layer of this region was analyzed using the geostatistics and GIS method, and its dominating factors, too, by means of variance analysis and regression analysis. Descriptive statistics of the results shows that the SOM content of the region varying in the range of 21.80 ± 7.43 g kg^-1 with a variation coefficient being 34.08%, both belonging to a moderate level. Geostatistical analysis suggests that the study area was very strong in spatial autocorrelation and structural factors played a dominating role in spatial variability of SOM, which was significant in anisotropy. The variation at 45^° in direction was the most severe. SOM in the area was distributed in a band, decreasing from northeast to southwest. Variance analysis and stepwise regression analysis indicates that mechanical composition of the soil was the dominating factor, which alone could explain 64.9% of the SOM spatial variability of the region, and land use, parent material and soil type followed in role. The four factors together could explain 74.6% of the variability of SOM in the study area.