Abstract: Particle swarm optimization (PSO) has been applied to many practical continuous optimization problems. However, it could not be extended to solve discrete and combinatorial optimization problems effectively. Traveling salesman problem (TSP) is a typical NP-hard problem. In view of the success of and GA in discrete and continuous optimization problems, this paper proposes a new Excellent Chromosome Fragments Genetic PSO (ECFG-PSO) for TSP. To prevent premature convergence of ECFG-PSO, local searching and the secondery initialization were added to the algorithm. The experiments show that the ECFG-PSO is superior to the DPSO existed in time efficiency and problem size.