The goal of reactive power dispatch is to minimize the system losses and improve the system voltage profiles at all times. This is achieved by adjusting various generating units’ excitation systems continuously, discrete tap positions of on-load tap changers of transformers as well as switching of correct doses of inductors or capacitors. This is a mixed integer non-linear optimization problem. In this paper, the differential evolution (DE), a novel evolutionary computation technique which was originally designed for continuous problems is applied to solve this problem; DE promises to ally qualities of established computational intelligence (CI) techniques with a more striking computational performance, thus advancing practical on line application in the control center. The developed tool was demonstrated on the Nigerian power system grid for three case scenarios preset on the power world simulator. The results achieved revealed that DE procured a significant reduction of real power losses while simultaneously keeping the voltage profiles within the acceptable limits.