Water scarcity, especially in Iran and during the recent droughts, emphasizes the importance of achieving an optimal operation policy for large dam reservoirs. In the last two decades, the annual optimization of dam reservoirs under the controlled conditions, as well as climatic and real conditions has attracted many researchers and engineers. In this paper, a new approach is proposed to predict the reservoir dam storage. The imperialist competitive algorithm (ICA) is a new approach in the field of evolutionary computation which calculates the optimal solution in different optimization problems. This algorithm, by mathematical modeling of the social-psychological evolution process, provides a new approach for solving mathematical optimization problems, and in comparison with other algorithms, has appropriate speed and high convergence rate in finding an optimal response. In this research, the imperialist competitive algorithm was used for annual optimization of the Kahir reservoir to derive the optimal policies. Objective function downstream water issue needs to establish relationships based on continuity were selected. Comparison of ICA model in population 100 showed that the ICA algorithm with average best objective function value of 125, 114.6 and 85.60 with a number of further evaluations of the objective function to achieve higher capacity is the optimum answer. The results showed a 6.1% error in the implementation of the ICA algorithm between the observed and predicted storages. The results of applying the imperialist competitive algorithm to the annual optimization problem indicate the capability of the proposed method.