Abstract: A method of using artificial neural networks for the working procedure selection in CAPP is proposed in this paper based on the trends of CAPP and the existing problems in CAPP expert system. By taking rotational parts as the object and combining with group technology, an integrated system consisting of many independent artificial neural networks is established. With the subsection codes of the geometric features, dimensions, tolerances, and roughness as the input, the codes of the working procedure selections as the output, the key technical problems applying neural network in working procedure seletion can effectively solved.