Abstract: This paper proposes a quantum genetic algorithm based on spheral solution space multi-region division. It introduces multi-region parallel searching mechanism with a permutation policy among sub-swarms, designs a novel quantum mutation operator using a variable probability decided by the extend of devolution. It is proved that this algorithm can converge to a global optimal solution with probability 1. The experiments of combinatorial and continuous optimization issue show that the proposed algorithm accelerates the convergence to the expected value smoothly, reducing the probability of premature. It performs well comparatively.