The study of ship collision in restricted waters is extremely important. Therefore, this study focuses on analysing the collision risk between fishing ships and container ships in the Qingdao port warning area, and proposes a comprehensive assessment model based on the risk indicator function and entropy value method. By preprocessing and analysing the AIS data, a risk model containing five key factors, namely DCPA, TCPA, relative distance, relative bearing and relative speed, is constructed. The results of the study show that fishing ships significantly increase the risk of collision during cross-encounters with container ships in the warning area, especially in the high-risk area close to the main channel, due to their high flexibility and their behaviour of not following the prescribed paths. The study not only reveals the collision risk characteristics in high-risk areas, but also provides an important basis for the optimisation of maritime traffic management and navigation safety warning systems.