Genetic Algorithms (GAS) are the most popular used methods of the evolutionary algorithm family. GA effectiveness highly depends on the choice of the search space range for each parameter to be optimized. The search space being a set of potential solutions may contain the global optimum and/or other local optimums. Being often a problem-based experience, a bad choice of search spaces will result in poor solutions. In this paper, a novel optimization approach based on GAs is proposed. It consists in moving the search space range during the optimization process toward promising areas that may contain the global optimum. This dynamic search space allows the GA to diversify its population with new solutions that are not available with fixed search space. As a result, the GA optimization performance can be improved in terms of solution quality and convergence rate. The proposed approach is applied to optimal design of multimachine power system stabilizers. A 16-machine, 68-bus power system is considered. The obtained results are evaluated and compared with other results obtained by ordinary GAs. Eigenvalue analysis and nonlinear system simulations demonstrate the effectiveness of the proposed approach in damping the electromechanical oscillations and enhancing the system dynamic stability. (c) 2012 Elsevier Ltd. All rights reserved.