Optimal design of water distribution systems has been studied extensively, assuming perfectly known parameters, resulting in deterministic optimization models. The results obtained by such models may perform poorly when implemented in the real world, when the problem parameters are revealed and different from those assumed in the deterministic model. In recent years, several new robust optimization methodologies have been developed incorporating the intrinsic uncertainty and providing robust solutions in terms of hydraulic reliability. In this work, a new non-probabilistic robust counterpart approach is proposed to optimize the design/rehabilitation of water distribution systems. The uncertainty of the information is described by a deterministic user-defined ellipsoidal uncertainty set, which can be probabilistically justified, and the decision maker searches for a solution that is optimal for all possible realizations of the uncertainty set. The robust counterpart makes no assumptions about the probability density function of the uncertain variables and their dependencies, does not require building a representative sample of scenarios, and has the same size as the original model. The robust counterpart is integrated with Cross Entropy optimization method for finding robust near-optimal solutions. The proposed methodology is demonstrated on two case studies. The results show considerable promise of the RC approach in terms of the tractability and the size of the model, as well as being able to show the trade-off between risk and cost.