Generalized cost recovery is one of the basic principles of the new Water Law that has recently been adopted by the Malagasy government. However, the effect of this change in policy is still poorly understood. Based on contingent valuation surveys in an urban and a rural area in southern Madagascar, this study analyzes the effect of changes in prices for water services. The results suggest that a minimum size of 90 households in a village is necessary to reach full cost recovery for well construction. Given that this is significantly above the current size of villages in the survey area, full cost recovery seems therefore impossible and subsidies are necessary to increase access to improved water services. Cost recovery for maintenance is relatively easier to achieve. In urban areas, water use practices and willingness to pay for water services depend highly on household income. To better serve the poor, it is therefore suggested that rich households, who rely on private taps, cross-subsidize poor households as a significant number of households is unwilling or unable to pay for water from a public tap. Given that public taps make up a small part of the total consumption of the national water company JIRAMA, lower income from public taps are shown to have only a marginal effect on its total income. However, as experiences in other countries as well as in Madagascar have shown, a fee on public taps is necessary as water for free leads to spoilage, does not give any incentive for the distributor to expand networks, and might therefore be a bad policy for the poor overall.