This paper deals with optimization of Water Distribution Networks (WDNs) by establishing District Metered Areas (DMAs). A methodological approach for the automatic partitioning of a WDN into DMAs is presented and discussed. Several design criteria are taken into account for the definition of the problem objectives and constraints. The optimization methodology applies adjustable weights to various objectives to allow flexible decisions to be made. The proposed approach combines graph theory and an evolutionary multi-objective algorithm, whose parameters were calibrated through extensive tests. The results of the application to a real case study are presented, as well.