Traveling Salesman Problem(TSP) is a classic combined optimization problem and it is proved that TSP is NP hard. A modified artificial immune algorithm is proposed to solve it. The algorithm simulates the protein polypeptide structure of the antibody, the clonal selection principle and the density regulation mechanism of the immune system, and uses a new analytic approach for the similarity between the antibodies. Moreover, the mutation operator is added the greed algorithm. Those progresses improve the search performance of the algorithm. The experiment results show that using this algorithm in TSP have better global search ability and faster convergence speed than using the standard genetic algorithm.