The forecast of water demand is important for Saudi Arabia which is characterized by a scarcity of its water supplies and a dependence on costly desalination plants to satisfy the water needs of its population. The forecasting task is even more challenging for tourist cities. The paper presents a probabilistic-based methodology for the forecast of future water demand for the city of Mecca in Saudi Arabia. Because of its religious nature, the city attracts visitors all year long. The large and variable number of visitors put considerable strains on the management of water supply especially that the city relies exclusively on desalination. Besides the random influx of visitors, the development of a sound forecast model is further complicated by the uncertainties associated with other key explanatory variables such as the economic activity, which is largely dependent on fluctuating oil prices. All these factors limit the usefulness of any deterministic forecast model. This paper develops a forecast model that incorporates explicitly the uncertainties associated with local population growth, tourist's influx, household size, household income as well as conservation measures. The methodology makes use of historic time series records of water consumption and applies Monte Carlo sampling to describe the associated uncertainties. (C) 2012 Elsevier B.V. All rights reserved.