This paper describes a two-part methodology for managing the risk posed by water supply variability to irrigated agriculture. First, an econometric model is used to explain the variation in the production value of irrigated agriculture. The explanatory variables include an index of irrigation water availability (surface storage levels), a price index representative of the crops grown in each geographical unit, and a time variable. The model corrects for autocorrelation and it is applied to 16 representative Spanish provinces in terms of irrigated agriculture. In the second part, the fitted models are used for the economic evaluation of drought risk. Inflow variability in the hydrological system servicing each province is used to perform ex-ante evaluations of economic output for the upcoming irrigation season. The model’s error and the probability distribution functions (PDFs) of the reservoirs’ storage variations are used to generate Monte Carlo (Latin Hypercube) simulations of agricultural output 7 and 3 months prior to the irrigation season. The results of these simulations illustrate the different risk profiles of each management unit, which depend on farm productivity and on the probability distribution function of water inflow to reservoirs. The potential for ex-ante drought impact assessments is demonstrated. By complementing hydrological models, this method can assist water managers and decision makers in managing reservoirs.