[Objective] The impact of seasonal drought on water use efficiency(WUE) of ecosystems and its carbon-water coupling mechanisms was analyzed in order to provide a scientific basis for enhancing vegetation carbon sequestration capacity, optimizing water resource allocation, and formulating ecosystem protection strategies. [Methods] Based on data from China's terrestrial ecosystems from 2001 to 2023, the spatiotemporal dynamics and seasonal patterns of WUE and its responses to drought were systematically assessed using trend and correlation analyses. By integrating the XGBoost-SHAP interpretable machine learning model with LOWESS fitting, the nonlinear response characteristics and threshold effects of drought stress factors were further quantified. [Results] The frequency of extreme drought events and the WUE of vegetation in China both exhibited pronounced seasonal patterns. Extreme droughts occurred most frequently and extensively in spring, while WUE reached its peak in summer. Its seasonal variations were jointly driven by gross primary productivity(GPP) and evapotranspiration in spring, gradually shifting to a ‘ GPP-dominated' pattern in summer and autumn, with a decreasing gradient from south to north. In spring, WUE was co-regulated by soil moisture and evapotranspiration, whereas in summer and autumn, it was primarily controlled by soil moisture. [Conclusion] The impact of drought on WUE exhibits significant seasonal differences and is primarily constrained by soil moisture conditions. Enhancing water management during the key controlling seasons and optimizing vegetation structure are essential strategies for improving ecosystem drought resilience and mitigating climate change and ecological risks.