Tolerance is a quantitative description of the precision in products and must be carefully considered when products are designed, manufactured and inspected. The exchangeability is an important principle in mass production, and tolerance is the key point to achieve the goal. In this thesis, we develop a tolerance simulation system, incorporating the worst-case model, statistical model and Monte Carlo model, for assembly tolerance analysis. This simulation provides a failure probability for the gap dimension which is involved in a dimensional chain. Additional values related to the dimensions and tolerance are displayed for further judgment and analysis. Moreover, a tolerance allocation algorithm is implemented to re-distribute the tolerance for some of the dimensions. Also, four areas of practical applications of the proposed system are indicated and examples are provided to demonstrate the feasibility of the proposed method.