An irrigation scheduling model is proposed for the operational control of irrigation systems. The model is based on the physical processes of soil-water dynamics and plant response; it is specifically developed for the optimal allocation of limited irrigation storage to offset the shortages in consumptive use requirements that result from marginal rainfall. Initially a distinction is made between the operational and planning problems of irrigation. The need for an optimal irrigation scheduling model in real-time irrigation control is discussed with reference to irrigation problems in Northern Nigeria where the climate is essentially semi-arid. A review of the quantitative methods for defining alternative water management policies for irrigation systems shows that a systems approach is necessary for dealing with operational problems. In devising a scheme for optimizing the irrigation regime, particular attention is paid to the submodels of the physical processes involved in irrigation optimization. The scheme involves simple models of: (1) soil-water dynamics in the root-zor.e; and (2) actual plant response. The soil-moisture dynamics model was tested against field data. Various formulations of single-stage and multi-stage crop production functions are reviewed. The irrigation scheduling model developed for controlling nonstationary irrigation systems is represented as a stochastic dynamic programming problem involving linear dynamics, a quadratic loss function and certainty equivalence. Performance of the control model is compared to typical field behaviour through computer simulation. Simulation results show that the irrigation scheduling model developed provides an improvement to current scheduling practice. Its application in a planning context is demonstrated using data relating to the Kano-Irrigation project in Northern Nigeria.