Ship trajectory clustering is an important method in the fields of maritime supervision and ship behavior analysis, aiming to extract ship trajectory patterns from the complex ship trajectories at sea and discover useful information. This is of great significance for the utilization of offshore ship channel resources. Although there have been many researches on trajectory clustering, these methods have not taken into account the course deviation factor of ships, and have not yet involved or achieved the ideal effect when applied to the ship trajectories, which can not solve the complexity problem caused by the large number of random ship trajectories at sea. In this paper, a ship AIS spatio-temporal trajectory clustering method based on course deviation is proposed. This method segments the compressed spatio-temporal trajectory lines based on course deviation, so that the ship trajectory lines with stable course have fewer feature points, while the ship trajectory lines with significant course changes retain relatively more feature points, which can better reflect the overall curvature characteristics of the original trajectory lines. Through experiments, we have verified the effectiveness of the proposed method.