Water supply system planning is concerned about the design of water supply infrastructure for distributing water from sources to users. Population growth, economic development and diminishing freshwater supplies are posing growing challenges for water supply system planning in many urban areas. Besides the need to exploit alternative water sources to the conventional surface and groundwater supplies, such as reclaimed water, a systematic point of view has to be taken for the efficient management of all potential water resources, so that issues of water supply, storage, treatment and reuse are not considered separately, but rather in the context of their interactions. The focus of this dissertation is to develop mathematical models and optimization algorithms for water supply system planning, where the interaction of different system components is explicitly considered. A deterministic nonlinear programming model is proposed at first to decide pipe and pump sizes in a regional water supply system for satisfying given potable and non-potable user demands over a certain planning horizon. A branch-and-bound algorithm based on the reformulation-linearization technique is then developed for solving the model to global optimality. To handle uncertainty in the planning process, a stochastic programming (SP) model and a robust optimization (RO) model are successively proposed to deal with random water supply and demand and the risk of facility failure, respectively. Both models attempt to make the decision of building some additional treatment and recharge facilities for recycling wastewater on-the-site. While the objective of the SP model is to minimize the total system design and expected operation cost, the RO model tries to achieve a favorable trade-off between system cost and system robustness, where the system robustness is defined in terms of meeting given user demands against the worst-case failure mode. The Benders decomposition method is then applied for solving both models by exploiting their special structure.