ABSTRACT: Human society in the Anthropocene is globally connected and relies on flows of goods and services for its prosperity and wellbeing. However, quantitative understanding of the flows of ecosystem service (ES) embedded in trade (virtual ES flow) across multiple human-natural systems remains limited. Here, we develop a framework to quantify virtual ES flows by integrating multi-region input–output modeling and ES mapping, and apply it focusing on water provisioning and climate regulation services, using China and its national-level urban agglomeration as an example. Our results showed that virtual flows of ES were substantially greater than the direct utilization of water and carbon resources, confirming the dependency on virtual flows of ES in highly urbanized regions. Interestingly, the virtual flows were mainly connected to distant rather than adjacent regions, highlighting the importance of considering cross-scale dynamics and managing long-distance flows in policy-making. Our developed framework has the potential to explore other types of ES and sustainability-related issues.