Abstract: In order to overcome the problems such as time-consuming, laborious, easy to cause animal stress and large error of artificial bull body measurement method, a method based on three-dimensional point cloud was used to measure the bull body size. Firstly, the 3D point cloud of the cattle body was collected by Kinect depth camera. After pretreatment, the discrete points of the point cloud were removed by statistical outlier method. Then the ground point cloud was removed by RANSAC method, and the background point cloud was removed by threshold cutting method. The minimum Z-value point column was obtained by slicing the bull point cloud, and the point column was fitted into a smooth curve. The key points of the bull body were obtained by combining with the characteristic region of the bull body. Finally, the key points of the body size associated with each body size parameter were calculated by Euclidean-style distance, and the error values were obtained by comparing with the body size parameters measured manually. The results showed that a large number of noise point clouds could be removed by preprocessing the point cloud, and only the cow point cloud is retained. By slicing the bovine point cloud, a more accurate minimum Z-value point column can be extracted, and the pseudo-synthesized curve can well represent the change of bovine profile. By combining the curvature changes of the curve with the changes of the bovine profile, the key points of the body size on the contour curve can be extracted. The errors of body oblique length, body height, body straight length, body depth, shoulder width and abdomen width were 1.10%, 3.09%, 0.99%, 5.97%, 9.36% and 6.18%, respectively. The results indicated that the measurement method based on 3D point cloud provided a feasible new way for the measurement of cattle body size.