Abstract: Joint operation of multi-reservoir systems is one of the most effective measures for direct control of regional water resources and it significantly increases the social, economic and ecological benefits compared with single reservoir operation. This paper describes a search space reduction method (SSRM) for solving hydroelectric optimal operation models of multi-reservoir systems, using the minimum output of a hydro plant and its discharge constraints to reduce the search space and overcome the shortcomings of the meta-heuristics in lower search efficiency and instable results. SSRM, when coupled with the meta-heuristics, can produce a high-quality initial population, and its search process is limited to an optimal search space that is closer to the feasible search domain. Thus, such reduction makes it easier to find the globally-optimized solution. We have adopted an effective algorithm, namely the cuckoo search (CS), in the optimization, and compared the traditional and reduced search spaces for a multi-reservoir system with given historical monthly inflows. A case study shows that SSRM, a simple and effective method for practical use, improves convergence and solution accuracy of the meta-heuristics.