A diversity guaranteed ant colony algorithm is presented by simulating the behavior of biological immune system. The algorithm adopts the immunogenic methods immune selection, immune memory, immune metabolism, density control and isolation niche technique. In each iteration of the algorithm, the solutions of the ants are selected to have crossover and mutation operations according to their quality and the distribution of the solutions. The mutation probability is determined by the diversity of the solutions. Experimental results on the traveling salesman problem show that our algorithm can obtain high quality of solutions, high convergence speed. It can avoid the stagnation and premature phenomena and has strong capability of optimization.