The paper reports on the use of the contingent valuation method to study the determinants of Swazi households' willingness to pay (WTP) for an improvement in their water quantity and quality. A sample of 374 households was surveyed and a Tobit model was applied to explain household preferences for quality and quantity of domestic water supply and derive estimates of WTP for such a service. The results confirm that household income had a positive and statistically significant impact on WTP for both quality and quantity. Distance to the water source is positively associated with WTP regardless of the location (rural or urban) and of the household head's age, education, and gender. Current water consumption was also statistically significant for WTP for improved quantity, but with a negative sign, implying that the more a household consumes water, the less that household is WTP to have improved water quantity. Conversely, the same household would be WTP for improved water quality. Rural households showed a much higher WTP for improved water provision services than urban households. There is therefore scope to improve water service levels in Swaziland even at a higher water price. More precisely, the estimates of WTP obtained in this study indicate the possibility of introducing a demand-driven program to expand the coverage of rural tap water schemes.