Accurate and detailed information of traffic accident scene is important for accident investigation, and the current investigation methods (tape measuring and total station survey) always need highway closure, and their working process is time-consuming. From different angles or at different altitudes, unmanned aerial vehicle (UAV) can monitor accident site without interrupting the traffic flow, therefore, UAV is introduced for accident scene reconstruction. Firstly, the method framework of accident scene reconstruction was proposed, in which UAV was used to take pictures of accident site, and imaging system was adopted to reconstruct the 2D and 3D accident scene. Then, 3D reconstruction, point cloud generation, and model optimization were presented. Next, a UAV flight experiment was conducted for traffic accident scene reconstruction, and two evaluation indexes, signal-to-noise ratio and structural similarity, were introduced to assess the image quality of accident scene reconstruction. The case study demonstrates that compared with current methods, the proposed method is efficient; moreover, the effect of accident scene reconstruction is satisfactory.