Natural causes (such as droughts), non-natural causes (such as competinguses), and government policies limit the supply of water for agriculture ingeneral and irrigating crops in particular. Under such reduced water supplyscenarios, existing physical models reduce irrigation proportionally amongcrops in the farmer's portfolio, disregarding temporal changes in economicand/or institutional conditions. Hence, changes in crop mix resulting fromexpectations about risks and returns are ignored. A method is developed thatconsiders those changes and accounts for economic substitution and expansioneffects. Forecasting studies based on this method with surface water inGeorgia and Alabama demonstrate the relative strength of econometricmodeling vis-à-vis physical methods. Results from a study using this methodfor ground water in Mississippi verify the robustness of those findings.Results from policy-induced simulation scenarios indicate water savings of12% to 27% using the innovative method developed. Although better irrigationwater demand forecasting in crop production was the key objective of thispilot project, conservation of a valuable natural resource (water) hasturned out to be a key consequence.