Abstract: The problem of land subsidence is relatively prominent in the Feng-Pei Plain of Jiangsu Province, yet research on its driving mechanisms remains scarce. This study integrates multi-source data, including the thickness of unconsolidated sediments, groundwater extraction intensity, groundwater levels and their variations in different aquifer groups, mining activities, and urban construction. A multi-scale geographically weighted regression (MGWR) model was employed to quantitatively analyze the spatiotemporal characteristics and driving mechanisms of land subsidence from 2017 to 2024. The results indicate that subsidence is mainly concentrated around the urban areas of Feng and Pei counties, and the northern part of Pei County. The area with cumulative subsidence exceeding 50 mm is 310 km², with a maximum subsidence of over 400 mm. Significant spatial autocorrelation is observed, with subsidence centers exhibiting a "high-high clustering" pattern. MGWR model results reveal that groundwater extraction and mining activities (x2-x9) are the primary factors driving land subsidence, followed by the influence of groundwater levels in the Lower Pleistocene of Neogene and the amplitude of water level changes in the Middle-Upper Pleistocene and the Lower Pleistocene of Neogene (x5-x7-x8). Conversely, the thickness of unconsolidated sediments, groundwater levels in the Holocene and Middle-Upper Pleistocene, water level amplitude in the Holocene, and building density (x1-x3-x4-x6-x10) do not show significant effects. Across the entire subsidence area, the five factors (x2-x5-x7-x8-x9) collectively explain 77.0% of the subsidence. The Geographical Detector (GD) model further confirmed the synergistic driving effects of groundwater and coal resource extraction and mid-deep water level changes on land subsidence. Compared with the classical GWR and OLS models, MGWR demonstrates superior performance in goodness of fit, model parsimony, and error control, more accurately capturing the spatial heterogeneity and multi-scale characteristics of different influencing factors. Based on the analysis of the spatial heterogeneity and intensity of the main influencing factors, as well as their interactions, an integrated prevention and control system of "monitoring-early warning, source control, and comprehensive management" is proposed. This provides a scientific basis and practical guidance for enhancing regional geological disaster prevention and ensuring the safety of the urban geological environment.