Soil moisture (SM) is a fundamental variable in terrestrial ecosystems, critically influencing hydrological cycles, ecological processes, and plant growth. Despite its importance, the complex interrelationships among vegetation, climate, and SM across different drought gradients are not yet fully understood. This study provides a systematic analysis of the spatiotemporal dynamics of surface (SMsurf) and root-zone soil moisture (SMroot) in China from 2001 to 2021 across different aridity gradient (Defined by the ratio of potential evaporation to precipitation). We employed an integrated analytical framework, combining Random Forest algorithms for driver importance assessment with Partial Least Squares Structural Equation Modeling (PLS-SEM) to elucidate the direct and indirect causal pathways. Our results reveal a significant nationwide increasing trend in both SMsurf and SMroot, with growth rates of 0.0013 and 0.0014 m3/m3/year, respectively. The most rapid increase occurred in semi-arid regions. Solar-Induced Chlorophyll Fluorescence (SIF) was identified as a significant positive driver of this growth. In contrast, vapor pressure deficit (VPD) was the primary climatic factor constraining SM at the national scale. A key finding is the stark contrast in dominant drivers across aridity gradients: temperature was the predominant factor controlling SM variations in arid and semi-arid regions, while VPD dominated in dry sub-humid and humid regions. The influence of potential evapotranspiration (Ep) shifted from positive in arid regions to negative in wetter zones. The PLS-SEM analysis further uncovered distinct mechanistic pathways: precipitation directly influenced SM in arid regions, whereas in semi-arid regions, wind speed and radiation mediated their effects indirectly through temperature. Overall, environmental and climatic factors primarily exerted their influence on SM by modulating vegetation greening, which served as a pivotal mediator. These findings elucidate the differential mechanisms governing vegetation-climate-SM interactions across aridity gradients. Our findings elucidate the complex and differential causal relationship and mechanism of the change in vegetation-climate-drought on soil moisture across different aridity gradients, providing the theoretical support for constructing sustainable vegetation restoration strategies for water resources and the effective utilization of soil moisture resources in arid and semi-arid regions in China.