An immune glowworm swarm optimization（GSO）algorithm was presented to optimize the load distribution of cold tandem rolling mill.By taking advantage of the strong sensitivity of antibodies to antigens in the immune algorithm,a large number of solutions with high adaptability in the feasible domain were rapidly generated under the excitation of the objective function,and the slow initial iteration of the basic artificial GSO algorithm caused by the low detection rate of high brightness glowworms within the observation range was solved.In order to improve the accuracy of the basic GSO algorithm and avoid trapping into local extremes,a new luciferin update formula was proposed based on the relationship between antibody concentration and excitation in the immune algorithm.The immune algorithm,artificial GSO algorithm and immune GSO algorithm were simulated and analyzed.The results show that the immune GSO algorithm is superior in convergence speed and accuracy.