The secondary voltage control problem of inverter-based islanded microgrid (MG) is proposed in this paper. Firstly, the dynamics of the distributed generation (DG) with primary control is analyzed and modeled. Then, a RBF-neural-network (RBF-NN) sliding-mode controller is designed to eliminate the voltage deviation caused by primary control. The output voltage of all DGs can restore to a reference value and remain stable. Finally, taking a MG consisting of four DGs as an example, experimental results validate the designed secondary control strategy.