To improve the reliability of an urban rail transit system under the background of networking and prevent the potential passenger flow risk of the system, from the perspective of complex network, the reliability of urban transit network and passenger flows weighted network were analyzed . For the purpose of measuring the impact of passenger flow on network reliability, the reliability evaluation indexes for the passenger flow weighted network were constructed. By utilizing the data of Shanghai public transport cards, the actual passenger flow weight of network nodes was obtained, which was then taken as a parameter to measure the network reliability. A dynamic model of Shanghai urban rail transit network was established based on the coupled map lattice (CML) model. The dynamic processes of cascading failures caused by network nodes under different situations were simulated and the changes on the reliability of the passenger flow weighted network during cascading failure processes were measured. The results demonstrate that the reliability of the passenger flow weighted network declines slightly during the initial steps. When the network is damaged more than half the scale, the reliability of weighted passenger flow approaches zero. As a result, it is concluded that actions taken during the initial steps will effectively prevent disturbances from spreading in the network.