As a wave-propelled and persistent unmanned surface vehicle, wave glider has been widely applied in marine monitoring. Due to its long voyage, the wave glider inevitably encounters various obstacles at sea, which may cause collision accidents. However, considering the characteristics of weak maneuverability and underactuation of it, the obstacle avoidance for the wave glider is challenging. In this paper, the dynamic obstacle avoidance strategy for the wave glider based on an improved artificial potential field (IAPF) is proposed. The IAPF is committed to addressing the local minimum in the traditional artificial potential field and the obstacle avoidance difficulties caused by the weak maneuverability of the wave glider. Various simulations are conducted to demonstrate the feasibility of the proposed strategy by adopting an eight-degree-of-freedom mathematical model of the wave glider. The simulation results show that the wave glider can accomplish the obstacle avoidance task with the proposed IAPF algorithm when facing with different dynamic obstacles under various marine environments.