This paper presents reconfigurable control system for ship course-changing and course-keeping. Adaptive neuro-fuzzy inference system (ANFIS) is used to identify a steering gear subsystem (actuator) in order to build an accurate reference model for the fault detection in this non-linear component. The proposed control system uses a compensator for reducing the loss of the control performance produced by faults in the steering gear subsystem. It is shown that the proposed control system is robust to faults in steering gear subsystem.