Amidst drastic environmental changes, the intricate interplay and feedback mechanisms in the water-vegetation-atmosphere nexus experience alteration. Previous research primarily centers on the responses among variables within this system, with little known about whether and how these responses (sensitivities) change. Here, we employ the Evapotranspiration Stress Index (ESI) to represent the equilibrium of the nexus and develop a memory dynamic linear model based on Bayesian forward filtering. The model takes into account the carry-over effect in the “dry gets drier” self-amplify loop, allowing for a more effective estimation of the ESI time-varying sensitivity to associated influencing factors. To corroborate the model, a 5-year moving window multiple linear regression is applied to estimate the approximate sensitivity fluctuations. Our analysis reveals that from 1950 to 2020, mainland China experienced a notable 4.74 % escalation in evapotranspiration stress. This is primarily attributed to surface soil moisture, whose sensitivity to ESI surged by 1.25-fold in the last decade compared to the early 2000s. Vapor Pressure Deficit (VPD) and Leaf Area Index (LAI) also exerted a substantial role, with their sensitivities fluctuating approximately 0.95 % and -0.56 %, respectively. Moreover, the greening pace is linked to an increase in soil moisture sensitivity and a decrease in VPD sensitivity, suggesting that rapid greening may alter the ecological resilience against soil deficit and atmospheric drought. Our findings underscore the spatiotemporal variations in sensitivity, enriching the comprehension of ecosystem reactions to external factors, and offer essential insights for refining Earth System Model parameters and advancing greening endeavors.