As irrigation water is an input into a production process, its demand must be ‘derived’. According to theory, a derived demand schedule should be downward sloping and dependent on the outputs produced from it, the prices of other inputs and the price of the water itself. Problems arise when an attempt is made to estimate the demand for irrigation water and the resulting own-price elasticity of demand, as the uses to which water is put are spatially, temporarily and geographically diverse. Because water is not generally freely traded, what normally passes for an estimate of the own-price elasticity of demand for irrigation water is usually a well argued assumption or an estimate that is derived from a simulation model of a hypothesized producer. Such approaches tend to provide an inadequate explanation of what is an extremely complex and important relationship. An adequate explanation of the relationship between the price and the quantity demanded of water should be one that not only accords with the theoretical expectations, but also accounts for the diversity of products produced from water (which includes the management practices of farmers), the seasons in which it is used and over the region within which it is used. The objective in this article is to present a method of estimating the demand curve for irrigation water. The method uses actual field data which is collated using the Residual Method to determine the value of the marginal product of water deployed over a wide range of crops, seasons and regions. These values of the marginal products, all which must lie of the input demand schedule for water, are then ordered from the highest value to the lowest. Then, the amount of irrigation water used for each product, in each season and in each region is cumulatively summed over the range of uses according to the order of the values of the marginal products. This data, once ordered, is then used to econometrically estimate the demand schedule from which the own-price elasticity of demand for irrigation water can be derived. To illustrate the method, the values of the marginal product of water deployed in the Musi catchment in India are used to determine an own-price elasticity of demand for irrigation water which has some positive value to producers of approximately -0.64. For water that is most highly valued, the elasticity was found to be highly elastic at -2.12, while less valued water used in agriculture was far more inelastic at -0.44. Finally, for almost 36% of water deployed in the catchment the elasticity was logically determined to be perfectly elastic.