Motivated by the realistic demand of reducing the flood risk of urban metro stations, this paper creatively proposes a research framework for flood risk classification assessment of metro stations. In the framework, a specified criterion system is established from the perspective of hazard, vulnerability, exposure and rescue capacity. Furthermore, a SMAA-2-FFS-H method is proposed, which is comprised of stochastic multi-criteria acceptability analysis-2 (SMAA-2), fuzzy flowsort (FFS) and multi-criteria hierarchy process (MCHP). For verifying the practicality and efficiency of the proposed framework, the flood risk evaluation of 32 stations in Zhengzhou Metro Line 5 during the "7·20" rainstorm event is taken as a case study. The results show that only Shakoulu Station is at high risk, which is correspond to the reality of the "7·20" rainstorm event. Moreover, the risk class characteristics of the remaining stations are correspond to historical flooding events. Among them, 10 metro stations are classified as medium risk, 21 metro stations are classified as low risk, and none metro station is classified as very low risk. As proved above all metro stations have a certain degree of flood risk, and the classes of flood risk are indeed different between some metro stations. In addition, the validity and robustness of the proposed method are demostrated by comparative analysis and sensitivity analysis. Finally, some managerial suggestions are provided for strengthening the flood control capacity of metro stations.