Abstract. Aiming at the problem of 3D collision avoidance for unmanned aerial vehicle (UAV), this paper proposed a UAV collision detection, collision decision and path re-planning method. Firstly, traditional methods calculate collision regions using certain threshold value of distance or time. In order to improve the shortcoming of existing methods, this paper considered maneuver information of both UAV and aerial intruder. Models of three-dimensional collision regions were presented. The proposed regions models could be used for UAV to successfully detect collision and make an appropriate collision decision. Secondly, a collision decision method based on the proposed regions and interfered fluid dynamical (IFDS) algorithm was present. UAV autonomously made collision decision by using the proposed regions models, and chose an appropriate maneuverer mode to avoid collision. Then the IFDS algorithm was applied to re-plan UAV's flight path. Finally, the simulation results demonstrate the effectiveness of the presented method.