A reactive power optimization model is established to realize the lowest active power loss of power system, and a quantum differential evolution (QDE) algorithm is proposed to solve the model. The algorithm takes advantage of parallel and collapse properties of the quantum calculation theory, owning better ergodic ability for the solution space. At the same time, possibility expression property of quantum bit in traditional selection strategy is considered which makes it easier to avoid premature phenomenon. The reactive power optimization for IEEE 30-bus system is performed by using traditional differential evolution algorithm, particle swarm optimization algorithm and quantum differential evolution algorithm. The comparison analysis of three algorithms shows that the quantum differential evolution algorithm is more scientific and efficient in dealing with reactive power optimization.