Particle swarm optimization has emerged as a useful optimization tool for handling nonlinear programming problems. Various modifications to the basic method have been proposed with a view to enhance speed and robustness and these have been applied successfully on some benchmark mathematical problems. But few applications have been reported on real-world problems such as economic dispatch (ED) with non smooth cost functions. To show improved particle swarm optimization (IPSO) in efficiency and effectiveness, the proposed IPSO is applied to test the economic dispatch problems, with non-smooth cost functions considering valve-point effects and multi-fuel problems. The results of the IPSO are compared with the results of conventional numerical methods, evolutionary programming approaches and genetic algorithm. Finally the convergence characteristics of improved particle swarm optimization (IPSO) are compared with modified particle swarm optimization results (MPSO).