In order to accurately identify the key stations and sections of urban rail transit, based on the improved betweenness method, and further considering the preference of passengers' travel path selection and the characteristics of transport flow in different stations and sections, a weighted improved betweenness was proposed to further distinguish the appropriate paths between all origin destination (OD) pairs in the rail transit network that met the given constraints. That is, when calculating the number of times that nodes or links appeared in all OD pairs of appropriate paths in the network, higher betweenness weights were assigned to the appropriate paths with shorter length and larger passenger flow, which overcame the defect that the improved betweenness method treats these appropriate paths equally. After that, a simulation case study was carried out with Beijing rail transit network as an example. The results show that the weighted improved betweenness method can provide new useful information for identifying the key stations and sections of urban rail transit network. In order to improve the accuracy of recognition, it is necessary to use traditional methods and weighted improved betweenness and other methods for comprehensive analysis. This suggests that the method makes the recognition result more authentic and credible.