This paper presents hierarchical solve-and-merge (HISAM): a two-stage approach to evolutionary multi-objective optimization. The first stage involves a simple genetic algorithm working on a number of isolated subpopulations, each using its own uniquely weighted linear scalarizing function to encourage it to focus on a different region of the Pareto space. At the second stage, the best solutions from stage one are passed to a Pareto-based hierarchy, where the solution set is judged on Pareto dominance and further improved. Preliminary results for large knapsack problems with 2-4 objectives are highly competitive with those obtained using other methods. Furthermore, the HISAM implementation has a fast execution time.