Given the natural variability and uncertainties of future water availability and demand, reliability is a critical factor for water supply system design. However, the complexity of water supply system and the correlated nature of uncertainties often cause system design models intractable. In this paper, correlated uncertainties in water demand and supply are applied on the form of the robust optimization approach of Bertsimas and Sim (2004) to design a reliable water supply system. Bertsimas and Sim (2004) has proposed a robust optimization approach to deal with data uncertainties and their correlation. This type of robust optimization was first introduced by Soyster (1973) for linear programming problems. However Soyster`s model constrains the objective function considerably to assure robustness. Several studies have modified the Soyster model to control the conservatism (Ben-Tal and Nemirovski, 1999; El-Ghaoui and Lebret, 1997; El-Ghaoui et al., 1998). This modifications introduced a higher degree of non-linearity into the system, thus the method is difficult to handle computationally. The approach of Bertsimas and Sim (2004) controls the degree of conservatism for the system reliability without increasing the difficulty in solving the original problem. In this paper, robust optimization technique of Bertsimas and Sim is applied for the water supply system of the Geum river basin in South Korea considering future uncertainty on water demand and availability.