A number of multi-objective genetic algorithms (MOGAs) have been developed to obtain Pareto optimal solutions for multi-objective optimization problems. However, as these methods involve probabilistic algorithms, there is no guarantee that the global search will be conducted in the design variable space. In such cases, there are unsearched areas in the design variable space, and the obtained Pareto solutions may not be truly optimal. In this paper, we propose an optimization method called NSDIRECT-GA to conduct a global search over the design variable space as much as possible, which improves the reliability of the obtained Pareto solutions. The effectiveness of NSDIRECT-GA was examined through numerical experiments. NSDIRECT-GA can obtain not only Pareto solutions, but also grasp the landscape of the search space, which results in higher reliability of the obtained solutions compared to MOGAs.