In this study, a model is developed for short-term flood control of a complex multi-reservoir system located on one of the largest basins in Turkey. The managing body of the reservoirs requires that short-term real-time flood control is used in conjunction with the long-term operation for energy maximization. For this purpose, a novel operating framework is developed to couple short-term and long-term decisions. The resulting optimization problems are a large scale and nonlinear. Hence, the solution is maintained by developing an improved epoch-based particle swarm optimization (PSO) method. The proposed solution is general and can be used for any watershed with different layouts of several reservoirs having anticipated inflows and pre-defined release and storage constraints. It is the first time PSO is used for such a complex large-scale problem within flood control context. The developed optimization model was applied to a sample basin having 10 reservoirs with general layout of the mixed structures. The results obtained were compared to the real operational data and it was found that a solution with less flooding and more energy production would have been possible if the proposed methodology were utilized.