A two-stage optimization procedure is developed. At the first stage the optimal values of irrigation area and installed power are determined subject to allowed deficits. The allowed deficits are given for irrigation and primary energy production. The demands for water supply are with high priority, and they are included in the constraints. At the second stage the problem of optimal control is solved applying dynamic programming. The algorithm at the first stage starts with maximal possible value for irrigation area S and installed power P. At the second stage the time series of inflow is used and the optimal control of reservoir is determined for given S and P from the first stage. The criterion for optimal control is maximum benefit. At the second stage S and P are modified subject to allowed deficit, and the second stage is run again, until the computed deficits become less or equal to the allowed. The optimization procedure may be terminated after a given number of iterations.