Keywords:Critical slowing down Resilience indicator Convex model Hydrological regime shift Watershed systems s u m m a r y Natural hydrological regimes are essential to the stability of river basins. While numerous efforts have been put forth to characterize flow regime alterations driven by climate change and human activities, few approaches have been proposed to explore changes in watershed resilience. The present study attempted to introduce a systematic approach that can be used to identify the resilience change of river basins based on annual river discharge through the application of a convex model and the principle of critical slowing down. Specifically, a resilience indicator (p i ) that reflects streamflow autocorrelation at a given time was proposed to represent the temporal variation of the system resilience, and annual water discharge at representative hydrological stations located in upstream, midstream, and downstream regions of a river basin was used to reflect alterations in long-term hydrological processes and the stability of river basins. The application of this method to the Yellow River basin and Yangtze River basin indicated that the system resilience was lower in downstream regions compared to upstream regions. The Yellow River basin has suffered a decrease in resilience in its lower reaches since 1971, which extended to the middle reaches in 1987 and upper reaches in 1990. Similarly, recent observation of the resilience change in the Yangtze River basin indicated that resilience in its lower reaches has likely decreased since 2002, and this low resilience extended to the middle reaches in 2005. Overall, our study presents a new method to predict potential decreases of resilience in complex large scale watershed systems where mechanistic insight is insufficient to build reliable basin-scale hydrological, climate, ecosystem integrated models.