The invention belongs to the field of measuring of the geometrical errors of machine tools and relates to a machine tool rotary shaft position error modeling and identifying method based on binocularvision. The machine tool rotary shaft position error modeling and identifying method comprises the steps that a binocular vision system is adopted, camera calibration is conducted, and a machine toolcoordinate system is established; a machine tool rotary shaft position error model at any angle is established, five encoded points on a positioning datum are used for representing machine tool motioninformation, the theoretical positions of the encoded points are compared with the measured actual positions, and a matrix equation containing error elements is obtained; and according to coupling term analysis, linearization of a solving equation is completed, optimized solving is conducted through the position and pose relation of the encoded points in the machine tool coordinate system and a pixel coordinate system, and the position error of a machine tool rotary shaft is modeled and identified. The error model established through the method is combined with the vision measuring means, sothat the problems that due to the capability limitation of measuring equipment, a model and identifying algorithm is complicated and the measuring process is tedious are solved effectively, operationis easy and convenient, and the equipment cost is low.