The author considers the question of problem solving using the divide-and-conquer approach. In particular, the appropriate choice of the size of the two subproblems is considered. The author contends that, “It is generally believed that the time-cost of solving any size n problem using two-way divide-and-conquer is minimized by balancing” and solving two subproblems of size n/2. The paper presents a counter-example related to finding the greatest and least element of a size n set in terms of cost-optimality. A necessary and sufficient condition for balancing to be cost-optimal is developed in the paper, and a new iterative merge-sorting algorithm is proposed.