Abstract: A new multi-objective immune algorithm based on multi-population is proposed by the use of genetic algorithms parallel mechanism to avoid local convergence.Each population interacts mutually information by the shift of excellent individual.This can effectively promote the diversity of population，and accelerate convergence rate.The performance comparison has been made between the new method and the traditional multi-objective evolutionary algorithm NSGA-II by using several typical testing functions in algorithms simulation experiment.Experimental results suggest that the proposed approach can resolve the problem of multi-objective effectively and has better performances.