Abstract: Based on the network model of physical topology layer, facility capability layer and passenger flow distribution layer of urban rail transit, a multi-attribute network is constructed. The feature indicators of network congestion-prone points, such as network topology, network transportation capacity, and network passenger flow are expounded, and combination weight method based on AHP-entropy weight method is adopted to identify and evaluate the congestion risk of congestion-prone points, and to explore correlation among network topology structure, transportation capacity and passenger flow demand. In the case of Beijing Rail Transit, most of the identified congestion-prone points are the transfer stations that connect suburban line and urban line. This verifies the validity and practicability of the identification and evaluation method.