Cities have been investing heavily in recent years to augment their water supply, focusing on either stormwater harvesting or desalination. A more optimal strategy is to consider all sources of water jointly, thereby allowing for hedging of supply risks. A portfolio model of urban water supply is derived, consisting of reservoirs, stormwater harvesting and de- salination, which takes into account the uncertainties of water flows and rainfall as well as the relative costs of associated with the three types of water sources. Calibrating the model to Melbourne's existing water supply system, a significant result is that optimal contributions are dependent on current water stocks, thereby providing insights into the appropiate- ness of historical water augmentation decisions. It is found that, with the exception of reservoirs, observed contributions deviate from optimal contributions for stormwater harvesting and desalination with the results suggesting a need for future investments to target stormwater harvesting ahead of desalination technology. Moreover, the optimal portfolio is found to vary throughout the year with desalination being the preferred supplementary supply source to reservoirs during summer months, while harvested stormwater is the preferred choice during winter. This result has implications for augmentation investments to mitigate seasonal water supply shortages.