Abstract: A new efficient cultural ant colony system (CACS) is proposed by integrating ant colony system into cultural algorithm frame. The computing model consists of a ACS-based population space and a currentbestsolution based belief space. Both the population space and the belief space have their own population, respectively, and evolve independently and parallel. The population space periodically contributes the currentbestsolution to the belief population, and the belief space continually evolves the currentbestsolution by using 2-OPT random mutation. The evolved solution is then used to update global pheromone level in the population space and guide the evolutionary search so as to improve population diversity and avoid prematurity. The contrasting experiments on the typical traveling salesman problem (TSP) show that the proposed algorithm is better than standard ant colony system in speed and accuracy.