Abstract: Objective: On days of largescale events taking place, URT (urban rail transit) stations in surrounding areas often experience instantaneous spike of inbound passenger flow. It is necessary to study the prediction methods for large passenger flow resulting from largescale events, so that transportation organization plans are promptly adjusted and foreseeable large passenger flow can be effectively handled. Method: The definition of URT large passenger flow is introduced. The structural, spatial, and temporal characteristics of the foreseeable large passenger flow induced by largescale events are analyzed, the prediction methods for which are presented, and their application is verified through practical examples. Result & Conclusion: Based on the predictability of large passenger flow, two categories of foreseeable and unforeseeable large passenger flows can be derived. Foreseeable large passenger flow from largescale events exhibits distinct structural, spatial, and temporal characteristics. The results of practical example verification indicate that the proposed prediction method based on STL (Seasonal and Trend decomposition using Loess) combined with LightGBM (also known as STLLightGBM) achieves high prediction accuracy. This method helps to fully understand the characteristics and patterns of large passenger flow induced by largescale events.