Based on the calibration technology of the distance error model, the model relationship between the distance error of the robot end and the kinematics parameter error of the robot was established, the transformation error of the coordinate system in the calibration process was avoided, and the calibration accuracy could be improved significantly. The vision measurement technology has the advantages of high precision, noncontact and realtime, compared with traditional methods, it has the advantages of low cost, easy operation. A robot calibration method which combined distance error model with vision measurement technology was proposed to improve the accuracy of industrial robots in specific workspace. A binocular vision system was adopted, and the camera was placed outside the robot for measurement. The robot parameters were calibrated based on the distance error model, and the kinematics parameters were compensated by using the calibration results. The results show that the distance errors are improved in a given calibration workspace; on the calibration trajectory, the average error of the absolute distance is reduced from 0.279 9 mm to 0.104 4 mm, and the error of the uncalibrated trajectory is decreased by more than 50%, by which the correctness and feasibility of the method are proved.