The provision of a water supply that is secure in the face of severe drought is a primary objective for urban water agencies – “running out of water” is not a viable option for a large city. However, there are other objectives that conflict with the primary one – these include minimizing costs and environmental impacts. A major challenge facing decision makers in the urban water sector is dealing with the trade-offs between these conflicting objectives. Multi-objective optimization methods have the potential to identify the optimal trade-offs between the competing objectives. The principal aim of this thesis is to address the shortcomings in existing multi-objective optimization applications to produce methods of greater practical relevance to urban water resource management. Review of past studies identified three practically significant shortcomings. Focusing exclusively on either long-term (or infrastructure) options or on short-term options such as operation rules may lead to sub-optimal solutions. The use of short climate forcing data time series in simulation models to evaluate drought security can produce solutions that make the system highly vulnerable to severe drought. Finally, the setting of a priori environmental constraints may hide trade-offs between environmental, economic and security factors that are of considerable interest to decision makers. These shortcomings are addressed by a new multi-objective methodology that exploits the ability of evolutionary algorithms to handle complex objective functions and simulation models. The principal novelty is the explicit treatment of drought security. A case study based on the headworks system for Australia’s largest city, Sydney, demonstrates the practical significance of these shortcomings and, importantly, the ability of the new approach to deal with these shortcomings in a practicable manner. In the face of urban population growth and the accompanying growth in water demand, the performance of the urban water resource system is expected to deteriorate over time. This will result in the need to intervene and adapt the system to the changing conditions. The scheduling capacity expansion problem seeks to identify the optimal schedule for the changes to the system. In past studies, this problem has been largely tackled by minimizing the total present worth of capital, operational and rationing costs. A significant drawback of minimizing the total present worth cost is that it is likely to produce solutions that lead to more severe and frequent rationing in the future. Such a solution is likely to be socially unacceptable. A multi-objective formulation for the scheduling capacity expansion problem is developed to overcome this shortcoming while addressing the need to explicitly deal with drought security and jointly optimize operating and infrastructure decisions. The formulation enables the trade-off between cost and equity (the equal sharing of the burden of restrictions over the planning horizon) to be explored. A case study based on the headworks system for Australia’s capital city, Canberra, demonstrates the advantages of the new approach. The optimization of urban water resource systems requires running simulation models tens of thousands of times. Given that simulation run times can range from less than a minute to thirty or more minutes, it is important to use a multi-objective optimization method which converges with the least number of evaluations (or simulations). To address this need, a detailed assessment is conducted of three benchmark multi-objective optimization methods and three newly developed methods based on ant colony optimization using case studies based on the Canberra and Sydney systems. No one method emerges as superior, although two of the six methods are identified as inferior.