To address the problem of minimizing the makespan for parallel batch process machines with non-identical machine capacities, non-identical jobs sizes and machine eligibility restriction, an effective ant colony optimization(ACO) algorithm is proposed. Based on the wasted space of the solution, the heuristic information is defined. Meanwhile, two candidate sets for constructing the solution are used to narrow the search space. To further enhance the solution quality, a local optimization approach is incorporated. The simulation results show that the proposed algorithm outperforms the other available algorithms.