To reduce the amount of water wastage caused by leakage, the utilities have to monitor and detect leakage of water distribution networks periodically. In order to identify leaking pipelines efficiently when limited resources are available, a cluster identification method (CIM) is proposed to establish a priority for leakage detection and to assess whether spatial clusters of high failure-prone areas exist. The proposed CIM evaluates the difference between the observed data and simulated trials to determine the statistical significance of each cluster; a method previously applied only in epidemiology studies to assess the occurrence probabilities of rare diseases for spatial clusters. The CIM suggested in this study is the overlapping local case proportions (OLCP) that uses grids to scope the entire area and then to simulate the number of failures in the neighborhood of each grid. The simulated failure ratios are then compared with the existing records to determine the statistical significance. The statistical significance represents the potential of the grid requiring further leakage detection. Three failure probability estimation methods, including local average, global average, and empirical equation, are utilized to analyze the suitability of the OLCP for use with various probability inputs. A case study in the central region of Taiwan was implemented to demonstrate the applicability of the proposed method. The results indicate that the rate of failure in the following year found within the spatial clusters determined by the OLCP was twice the average amount and thus provided valuable information used to prioritize the pipelines for further inspection.