This paper deals with the development of a procedure that integrates deterministic and stochastic optimization models for the real-time operation of a multipurpose reservoir. The procedure first determines the long-term (monthly) operating policy and then employs its information to search for the best short-term (daily) operation. Monthly and daily inflows are assumed to be stochastic and deterministic, respectively. The efficacy of the approach is investigated in a case study of the reservoir that supplies the city of Matsuyama, Japan. The results indicate that the procedure is able to provide more superior operating policies than standard simulation techniques.