The paper investigates the recovery problem for the multiple autonomous underwater vehicles (AUV) system via QPSO-based sliding mode control. A novel sliding mode control (SMC) technique is proposed based on the local communication among neighbors so that the multiple AUVs can track the mothership ultimately, and completes recovery task. By means of quantum particle swarm optimization (QPSO), a QPSO-based SMC strategy is further designed to select the optimal parameter set of controller, and decrease convergence error and control consumption. Simulation results show that the multiple AUVs recovery system can achieve consensus with the proposed coordinated control protocol.