This paper describes the development of an on-line intelligent multi-input-multi-output autopilot for simultaneously controlling the yaw and roll dynamics of an autonomous underwater vehicle (AUV) model. The six-degree-of-freedom non-linear AUV model used in the design study is hosted in the MATLAB/Simulink environment. Simulation results are presented that demonstrate the ability of the autopilot to operate effectively in the presence of mass variation in the vehicle model and measurement noise. Thus, the approach adopted offers an attractive alternative to more traditional methodologies.