To deal with the multi-objective flow shop scheduling with limited buffer, a Pareto-based general multi-objective glowworm swarm optimization(GMOGSO) algorithm is proposed. The concepts of the swap operator and swap sequence are introduced to make the continuous GSO discrete. To balance the convergence speed and accuracy, the GSO algorithm is developed into two different processes including the global optimization one and local optimization one. The global optimization process is used to improve the quality of the initial population, then the predatory search strategy is used to coordinate local and global exploration. The method of variant step length enhances the exploratory capability. The proposed GMOGSO algorithm is compared to other algorithms, and the results show the effectiveness of the proposed algorithm.