This study investigates the applicability of ordinal optimization approach to optimal control of power distribution systems. This approach enables us searching for a solution good enough with high probability instead of the best for sure in solving those problems with huge search spaces. Power distribution system control is such a problem, to which a near-optimal solution can be found by evolutionary search techniques. In this study we calculated node voltages and power losses of a sample system corresponding to all possible OLTC settings and capacitor on/off statuses. We obtained these solutions exactly and approximately and investigated the coincidences between the best 1%, the best 5%, and the best 10% of these solutions correspondingly. Results show that based on approximately-calculated node voltages and power losses a near-optimal solution can still be found to power distribution system control problem by an evolutionary search technique with considerable saving in total computation time.