A consistent approach to the development of tuning rules for course-keeping and path-tracking PID controllers for a ship autopilot are presented. The consistency comes from the observation that for each of the controllers the controlled plant can be modelled by an integrator with inertia. In the case of the course controller, it is the well-known Nomoto model. The PID controller may be implemented in series or parallel form, the consequence of which is a 2nd or 3rd order of the system, specified by a double or triple closed-loop time constant. The new tuning rules may be an alternative to the standard ones given in [1,2]. It is shown that, whereas the reference responses for the standard and new rules are almost the same, the new rules provide better suppression of disturbances such as wind, waves or current. The parallel controller is particularly advantageous. The path-tracking PID controller can provide better tracking accuracy than the conventional PI. Simulated path-tracking trajectories generated by a cascade control system are presented. The novelty of this research is in the theory, specifically in the development of new tuning rules for the two PID autopilot controllers that improve disturbance suppression.