Fault diagnostic technique of the industrial robot will be greatly needed in the various automatic systems. When the robot moves irregularly, it must be generally done that the movement of robot should be stopped as soon as possible as one of the safety factor. In this study, one fault diagnostic technique of robot motion is proposed to resolve the above mentioned problem. It is the proposed technique in this study that the small vibration of robot arm that is caused by the operation of robot is always been monitoring and the decision whether the robot moves regularly or irregularly is done statistically compared with the regular small vibration of robot arm. Due to the experimental results of fault diagnosis of robot motion, it is found that χ2 inference technique about the amplitude distribution of small vibration is good fault diagnostic technique of robot motion.