There is a lack of standardized water leakage statistics for water leakage in existing urban rail transit, which makes the quantified statistical outcomes challenging. Therefore, a fuzzy comprehensive evaluation system for water leakage in existing urban rail transit stations and tunnels is established using the analytic hierarchy process (AHP). Additionally, the water leakage classification standards are determined based on measured data and relevant specifications. The weight of each evaluation index is calculated by employing a combination weighting method that integrates AHP, sequential relationship analysis, entropy weight method, and CRITIC method. The calculation results show that the water leakage level is the most important weight. This system provides valuable reference for practical engineering water leakage treatment plans. The classification of water leakage stations and tunnels is ultimately achieved through a systematic clustering method, thereby providing valuable insights for prioritizing water leakage treatment. Finally, the practicability of the established system is validated through engineering examples.