The novel string structure for solving the economic dispatch problems through genetic algorithm (GA) is presented. The proposed string scheme can effectively prevent getting infeasible solution when evolutionary computation methods are preceded. Therefore, it can increase the searching efficiency and also solution quality dramatically. Many nonlinear characteristics of the power generators and practical constraints, such as generation limitation, ramp rate limits, prohibited operating zone, transmission loss and nonlinear cost functions are all considered for realistic operation. The effectiveness and feasibility of the proposed method are demonstrated by three real systems, and compared with the literatures method in terms of the solution quality and computation efficiency. The experimental shows encouraging results that the proposed approach was indeed capable of obtaining higher quality and efficiency solutions in economic dispatch problems. (c) 2008 Elsevier Ltd. All rights reserved.