Abstract: Mechanism path synthesis can be viewed as a pattern matching problem in essence.A pattern matching model for mechanism path synthesis is presented,and some key techniques related to its implementation such as digital description of mechanism path,clustering and classification of path curves as well as the path pattern matching are discussed in detail.Approaches of immune computing,namely the AINE model for unsupervised learning,AIRS model for supervised learning and the improved model of negative selection algorithm are exploited in an integrated way.A practical example of four-bar linkage path synthesis demonstrates that the proposed method is effective.