Water leakage in metro systems poses a persistent threat to structural durability and operational safety, particularly in water-bearing environments where leakage-induced deterioration may propagate through interconnected stations. While previous studies have extensively investigated leakage mechanisms and local structural responses, limited attention has been paid to the system-level vulnerability of metro networks under leakage disturbances, particularly the lack of an integrated framework linking leakage susceptibility and network characteristics. To address this gap, this study aims to develop an integrated framework for assessing station-level vulnerability in metro networks by incorporating both network characteristics and leakage susceptibility factors. Methodologically, network properties are first quantified by integrating topological structure and passenger flow characteristics. Leakage susceptibility is then evaluated using a fuzzy comprehensive evaluation method based on field investigation data. A combined weighting approach is further employed to integrate network and leakage indicators into a unified vulnerability assessment framework. Finally, a Monte Carlo probabilistic failure model is introduced to evaluate system robustness, and station vulnerability is ranked using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method. The results indicate that leakage-related factors contribute dominantly to station vulnerability, accounting for approximately 62.4% of the overall weight, highlighting their critical role in metro system performance degradation. Several stations are identified as high-risk nodes due to the combined effects of unfavorable hydrogeological conditions and topological importance. The proposed framework can support infrastructure managers in prioritizing inspection scheduling, preventive maintenance, and targeted reinforcement, thereby enhancing the resilience of metro systems against leakage-induced disruptions. Future work should incorporate multi-temporal operational data and real-time monitoring information, and further validate the model using long-term maintenance records to improve its practical applicability.