Hovering is a common operation condition for underwater vehicles, and the S surface algorithm has been proven to be a pragmatic approach to solve the control problem, however, with poor adaptability under changing environments. In this paper, an adaptive method based on fuzzy rules is proposed to achieve on-line setting of the controller’s parameters according to the relevant theories and experience. Meanwhile, a strategy to adaptively reduce the steady state error during the hovering is put forward. The adaptive S surface controller is test on a full-actuated underwater vehicle, and the pool experiment indicates that the modified controller ensured control accuracy and strong robustness.