Abstract: With ArcGIS, GeoDa, and Matlab, this paper studies the spatial pattern of Beijing agricultural soil heavy metals using complex network theory. Taking 1 018 soil samples as network nodes, and the spatial autocorrelation of soil heavy metals as edges connection reference, we create networks of Cr、Ni、Zn、Hg. And taking network cluster of Hg for example, we calculate network characteristics, such as degree, betweenness, average shortest path, etc. The conclusions of the research are as follows: 1) according to networks of Cr、Ni、Zn、Hg, nodes of high degree value only accounting for about 30%. These nodes affect more regional soil heavy metals around, and have formed stable network structures, which is difficult to prevent and control soil pollution. 2) Hg network cluster is a network of small-world characteristics, but not a scale-free. Nodes with high betweenness, undertaking the main network traffic, are concentrated, and distributed along with the Chaobai river. It's indicating that river play an important role in network formation. 3) Land uses, to a certain, impact the distribution of Hg network cluster. Human activities brought a lot of agricultural and mining sites, which result in Hg accumulation at north, east, and south of the core urban area of Beijing. The research can provide a reference for further analysis of the spatial pattern of soil heavy metal pollution and diffusion evolutionary mechanisms.