This research proposes a seismic optimization design method for water distribution networks under multiple constraints. Through an analysis of real water distribution networks from the perspective of network science, the regularity of their topological characteristics is revealed. Specifically, the product of average closeness centrality and average distance, the product of global efficiency and average distance, the ratio of average distance to effective diameter, and the distribution of node degrees exhibit specific value patterns. Taking water distribution network’s annual reduced cost as optimization object, along with the seismic reliability and the network’s topological characteristics as constraints, a network topology optimization model is established. Based on automatic generation strategy, genetic algorithms are employed to solve the seismic optimization model. The results indicate the optimized network meets the requirements for seismic performance and exhibits closer resemblance to real networks in terms of topological characteristics.