A fuzzy autopilot algorithm for maneuvering of surface ships is proposed and its performance studied. A mathematical model for nonlinear maneuvering of cargo ships has been used in conjunction with the proposed fuzzy autopilot. The autopilot works with two error measures that are inputs, error in heading angle and the offset error from the desired path, and has the command rudder angle as the output. Input variable fuzzification, fuzzy associative memory rules and output set defuzzification schemes using Mamdani minimum inference method are described. The proposed control scheme has been verified for a variety of straight line and curved trajectories using a nonlinear maneuvering model of Mariner class vessel. © 2011 - IOS Press and the authors. All rights reserved.