This paper deals with the development and application of an optimization model which combines nonlinear programming and genetic algorithms for the real-time operation of a water resource system composed by a multipurpose reservoir and a set of wells. The system is responsible for the water supply of Matsuyama City, located in Ehime Prefecture, Japan. The aim is to determine the optimal daily allocation of water from the reservoir and wells for all the uses according to the forecast of rainfall during a determined operating horizon. For comparison purposes, the model is also solved by standard optimization techniques but the results show that genetic algorithms are more efficient in finding better solutions.