In this paper, the reliability of a system to provide municipal water under risks of water quality and quantity is considered. We focus on two contexts of the system; one is a trend to introduce market mechanisms including some forms of “privatizations” because of pressure from public finance, and the other is the need to invest in maintenance and renewal of deteriorated water supply facilities that require drastic reform of the operations of local governments in the near future. Many public services must establish supply systems that stabilize their services consistent with an efficient management system. In the reform of municipal water supply systems, management of both quality and quantity risks is one of the most important concerns.<br> We investigated three main issues. First, we examined the quality and quantity risks and the current institutional measures for the provision of municipal water. Second, the present supply system was studied from the view of who is in charge of the risk management. Third, we exhibited “the reliability evaluation model”, which shows the marginal benefit of an increase in reliability is equivalent to the marginal cost of the project for risk management. The model can show the optimal reliability level or the optimal investment level to achieve a certain reliability level.<br> The welfare measure of the reliability benefit is, however, classified into several types. The proper welfare measures for change in risk (reliability) of risk-averse individuals depends on three conditions; the availability of conditional markets for redistribution of the risk, whether the project cost is certain or conditional, and whether the risk the economic agents face is a collective or individual risk. The appropriate measures, either option price, ex-post compensation, fair bet point, or their expected values, were assigned to combinations of each case. We classified some of the measures for water supply risk into each welfare measure in line with these conditions and note that the precise classification and identification of proper measurements is left for future empirical analysis.<br> This work was supported by KAKENHI (20730188).<br><br></i>JEL classification: H41, H43, H76, Q51, Q53