[Purposes] This paper aims to investigate the variation patterns of centrality of subway networks under different spatiotemporal distributions of passenger flows. [Methods] In this study, a two-step clustering algorithm was employed to partition automatic fare collection (AFC) data of urban rail transit into distinct time periods, and the corresponding passenger flows were incorporated into the subway′s topological network, which thereby generated a time-varying passenger flow weighted network. From the perspective of passenger transportation, the centrality of subway networks was examined using three indicators: station service intensity, station passenger flow betweenness, and station externality. The Shanghai subway network was used as a case study, and three typical passenger flow periods were identified: peak, transitional, and off-peak periods. [Findings] Subway stations exhibit higher service intensity and externality during the peak period compared to the off-peak and transitional periods. However, there is no significant difference in passenger flow betweenness among different time periods. Stations with higher externality are found not only in the central area of the subway network but also along the peripheral branch lines. Moreover, the top ten stations in each indicator for different time periods show a high degree of overlap, and all are concentrated on the three earliest subway lines. [Conclusions] The findings of this research can provide valuable insights for the daily organization of passenger flows for urban rail transit and the planning and implementation of safety contingency plans for unforeseen events.