As an essential link between sheet metal design and manufacturing, process planning generates a sequenced set of instructions to manufacture parts. However, according to a literature survey in this thesis, the tolerance transfer issue in sheet metal forming are insufficiently addressed: machining errors and their causes are not presented comprehensively as the sources of final error; only size dimensional tolerances are discussed in detail; computer aided tolerancing for parts formed by a combination of operations such as bending, punching, and blanking, is seldom studied; statistical tolerancing are utilized only for sheet metal assembly issues or size dimensional tolerances, not geometric tolerances. Therefore, the presented research systematically explores and proposes three dimensional geometrical tolerance transfer approaches for sheet metal part forming. The here proposed models are suitable for worse case or statistical analysis (using the Monte-Carlo methods). First, an integrated CAD/CAPP system based on feature evolvement, associative features, and data association mechanism, is outlined. Then, a mathematical model of geometric tolerance transfer is proposed. Finally, a machining error-correlated cost model is presented and applied to tolerance synthesis for a parallelism tolerance. Angular errors in the estimation of spring back are demonstrated to affect final errors considerably. The proposed tolerance transfer method is applied to parts formed by bending (parallel and non-parallel) and punching operations jointly. It can, subject to minor modifications, be easily applied to all bending-like and cutting operations and extended to other sheet metal forming operations. Monte-Carlo simulations show that accumulated tolerances may not closely follow a normal distribution. The research work is a solid basis for future work in this area. i Firstly I would like to express my heartfelt gratitude especially to Dr. Georg Lothar Thimm and Dr. Ma Yongsheng, for their kind guidance and supervision during the planning and writing of this thesis. Without their help this report could not have been completed. Then I want to thank Prof Khoo Li Pheng, Associate Porf Seet Gim Lee, Associate Professor Chen Chun-Hsien, and Assistant Prof Au Chi Kit for their help and good suggestions for my research project. I would also like to extend my appreciation to my friends, Miss Xu, Cheng Gang, Tang Shaohui, technical staff of ADaM lab, and all other friends for their help. At last, I want to thank my dear parents and my young sister, for unconditionally providing their love, support, and encouragement.