Stochastic mixed-integer optimization is used to identify a portfolio of long- and short-term supply and conservation actions for a municipal water system to cost-effectively accommodate a distribution of water shortages. Alternative robust, grey-number, and best/worst case formulations systematically explore implications of uncertainties in action costs, life spans, water volumes gained or saved, shortage levels, and shortage probabilities. A detailed example for Amman, Jordan considers 23 potential actions. Results show: (1) remarkable consistency occurs across the different modeling approaches. (2) Conserving water—reducing leakage and targeting select customers to install water efficient appliances—plays an important and growing role over time. (3) A delayed need for large supply projects like pumping the Disi aquifer. (4) No role appears for seawater desalination (Red–Dead Canal) before 2040. (5) Desalinating brackish Zara-Ma’een water is the low-cost option to increase water availability to customers but requires substantial capital investments. And (6) two shortcomings arise for grey-number and best/worst case approaches.