Vegetation resilience is fundamental to maintaining structural and functional stability, which is susceptible to natural and human activity in drylands. However, significant knowledge gaps remain in the underlying mechanisms of vegetation resilience across different dryland ecosystems. We used the lag-1 autocorrelation coefficient to assess the vegetation resilience in China's drylands. Furthermore, the Extreme Gradient Boosting (XGBoost) algorithm, Shapley Additive exPlanations (SHAP), and generalized additive models (GAMs) were applied to investigate the underlying driving mechanisms and determine critical thresholds. The results indicated most regions experienced either a continuously decreasing trend (27.63%) or an increase followed by a decreasing trend (32.17%). Climatic factors had the greatest influence, with lower temperatures being critical for maintaining high resilience in the arid (<2.27 degrees C) and semi-arid (<5.22 degrees C) subregions. Additionally, low precipitation variability (<12.41 mm) and vapor pressure deficit variability (<0.24 kPa) significantly enhanced vegetation resilience in the arid and semi-arid subregions, respectively. High soil cation exchange capacity (>14.75 me/100 g), low temperature (<1.2 degrees C), and moderate elevation (1698.52-5294.30 m) were conducive to sustaining high vegetation resilience in the dry sub-humid subregion. The findings provide a scientific basis for developing effective vegetation restoration and ecological conservation strategies in similar zones worldwide.