Optimal design of sewer network has conventionally been pursued within a deterministic framework. However, such design may be appropriate because of existence of uncertainties affecting the performance of a storm sewer system, e.g., precipitation rate, insufficient data, errors in design equations, and other factors. These uncertainties are systematically evaluated and incorporated in the proposed design of a storm sewer system. The optimal choice of slopes and diameters of the entire sewer network is based on a tradeoff between risk due to potential flood losses and the cost of installation of the sewers. Discrete differential dynamic programming technique is used in the risk-based design.