Land subsidence induced by long-term excessive groundwater withdrawal has constituted a major geological disasters in Beijing plain. This paper is based on the deficiency of present leveling network, selected three kinds of influence factors, including the partition of hydrology geology unit, the groundwater level decreased rate of main exploitation layers and the total thickness of compressed layers. Based on GIS spatial analysis platform, the partition map of comprehensive affecting factors of Beijing plain is created, and divided into 448 partitions. On this basis, we optimized design the land subsidence leveling network, added 220 leveling points. Using geostatistical Kriging interpolation error variance as the accuracy evaluation index, we drew the standard deviation distribution contour map of land subsidence leveling network, found that the standard deviation of the optimized is lower than before. Proved that based on the background of geological environment, using the partition map of comprehensive affecting factors to optimize design the land subsidence leveling network is feasible. This method can be used as the basis of construction of leveling points.