n order to solve the forecast problem of the product assembly accuracy and assembly success rate in the product design phase, in view of the assembly model, using the mixed programming technology of VS2010 and Matlab, based on tolerance analysis theory, the prediction system of assembly success rate is developed based on the Creo secondary development. The feature extraction of assembly constraint information, materials, shape characteristics, dimensional accuracy and geometrical accuracy, surface accuracy was completed, and then the tolerance analysis of assembly based on the extreme value method, probability method, Monte Carlo method was established, achieving the assembly precision calculation and prediction of assembly success rate about the assembly model. Finally taking a gear assembly as the research object, the system was tested. The results show that the system can realize product assembly success rate prediction conveniently and accurately.