Abstract: As general stochastic search algorithms, geneitc algorithms (GA's) have been applied to a variety of research fields such as function optimization, machine learning, etc.. Classical GA's can not surmount the ordering problems well. In this paper, for a kind of classical ordering problems——the traveling salesman problem, we first illustrate the necessity of using normalization techniques to calculate appropriate evaluation functions and then design a new heuristic crossover operator which is more effective than similar operators published before.