This dissertation addresses the challenge of supplying water to rapidly growing cities in South Asia, using evidence from the water-scarce city of Chennai. Chennai (formerly Madras) is a rapidly growing metropolis of over 6.5 million people, whose infrastructure has not kept pace with its growing demand for water. In the year 2003-2004, Chennai experienced a severe water crisis: the piped supply for the entire city was virtually shut down for a 12-month period. Consumers became dependent on private tanker suppliers trucking in untreated groundwater from peri-urban areas. This research effort accomplished three goals: understanding the dynamics of the recent water crisis, extending the model to project the business-as-usual trajectory of Chennai's water supply and understanding how the trajectory may be altered by various policies. The study departs from previous research studies in several respects: Firstly, this study explicitly incorporates self-supply via private wells, and private-supply via the tanker market as an integral part of the urban water system. Secondly, the research integrates bio-physical and socio-economic behavior at multiple scales: user-scale supply and demand, utility-scale management, and basin-scale water availability and allocation. Finally, the study allows policy-makers to evaluate and compare a wide-range of policy options on an apples-to-apples basis, something that cannot be done with existing frameworks. An integrative theoretical framework and model were developed to address the research goals. The integrated model was calibrated for the historical period 2002-2006 against extensive physical and socioeconomic data: groundwater heads, reservoir levels, household survey data in dry and wet years, tanker surveys, and operational statistics collected from the water utility. The calibration run of the model suggests that the 2003-2004 water crisis was precipitated by rational responses of the utility and Chennai consumers to limited reservoir capacity, unreliable inter-state water transfers, and limited capacity of the local aquifer. The research also explored scenarios of what the city's water supply may look like in 2025, using reasonable projections of population, land use and income growth. The historical rainfall record was used to generate scenarios of future rainfall. The 2025 model simulation provides two key insights. Firstly, a future drought is likely to at least as severe as the historical one. Increases in water use due to rising populations and incomes more than compensate for any reductions in peri-urban agricultural water extractions caused by to expanding urbanization. Second, a "dual-quality" approach to urban water supply may address Chennai's water problems. The dual-quality solution involves relying on centralized high-quality (and cost) supply for drinking, cooking and dishwashing while using lower quality (and cost) self-supplied groundwater for other non-potable needs. The research indicates that several factors contribute to making the dual-quality solution optimal. In the absence of reliable inter-state deliveries and a local perennial source, the long-run marginal cost of utility supply in Chennai is desalination, a very expensive option. Furthermore, a vast majority of consumers already have private wells; so consumers only consider the pumping costs of extracting groundwater from their wells; the capital costs are sunk costs. So, if in order to achieve full-cost recovery, the utility raises its tariffs above the cost of groundwater extraction from wells, rational consumers would switch out of using utility supply except for uses that necessitate high-quality piped water. The...