For a finite or infinite planning horizon, an interseasonal model is employed to determine an irrigation policy in terms of leaching and seasonal irrigation depths. The interseasonal model employs stochastic dynamic programming and simulation to maximize gross margin over the planning horizon. Solutions to the model are obtained by value iteration and policy iteration methods for a finite and infinite planning horizons, respectively, with researched crop characteristics and measured climatic parameters from Davis, California. Salinity in the root zone of plants is defined as the state variable. A stage is a year commencing at the end of the rainy season. Seasonal rainfall is treated as a stochastic variable. The information provided by the interseasonal model is used in an area‐allocation model that allocates acreage available for planting among the crops for unlimited and limited water supplies. Copyright 1979 by the American Geophysical Union.