Abstract: The inner transmission accuracies of the arc welding robots would cause cumulative errors on the motion paths of welding gun, and which inevitably causes low welding quality.Therefore, it was a key problem how to control the transmission accuracies and different random errors of the welding robots in improving the original designqualities of welding robot. A motion accuracy reliability analysis method was introducedfor the arc welding robot with ADAMS/View software. A Monte Carlo parametric virtue prototype of the robot was built and its different error distributions were simulated by the pseudo-random number sub-function program by using multiplicative congruential method, and then a large sample data in the dynamic welding process were obtained. The robot motion reliabilities can be calculated based on these data. It showed that the advantages of this method include simplicity(no need to solve complex dynamic mathematical models), high accuracy, and efficiency.