This thesis proposed a novel integrated Similarity Assessment Method (SAM) particularly suitable for the 3D shape design re-usage domain, and focused on the shape descriptor representation, algorithm development, similarity computation, and shape query interface. The popularity of CAD/CAM technique in product design and manufacturing industry has resulted in a large number of 3D shapes being generated. This provides the possibility to use customized design by applying shape re-usage technology, where existing shapes are retrieved and redesigned to obtain a new product. Shape re-usage could result in a marked increase in product variability to fulfill various customer demands, while not causes a significant increase in design and manufacturing cost. One of the key points of implementing shape re-usage technology is how to efficiently assess the similarity levels between the newly designed model and customer’s intention. SAMs for 3D shapes have been studied for decades, and well documented in literatures. However, the SAMs designed particularly for the 3D shape re-usage domain are rare. To explore a suitable SAM on the 3D shape re-usage domain, firstly, a survey on the current content based SAMs for 3D models was carried out. Secondly, University of Macau Database of Shoes shapes (UMDS) was also built and organized into different categories of re-used features, in addition to the Princeton shape benchmarking (PSB), for the purpose of providing basic platform for further experiments. It follows a problem analysis, where the functional requirements and specific characteristics of 3D shape re-usage were investigated and matched to prevailing similarity measures. The matched SAMs were implemented, and their algorithm performances were verified via a preliminary experiment. The experiment methods, including the common indicators “precision-recall diagram (P-R)”, “Express classification case study”, and a new indicator “Average errors in top 9 results (E9)”, were adopted. The result of preliminary experiment identified three appropriate SAMs. Distance Shape Histogram (DSH) method shows an optimal performance on discriminating shape categorization, however is insensitive on detail comparison for shapes in the same category. On the contrary, Solid Angle (SA) method displays a higher accuracy on identifying shapes within single category than shapes across multiply categorizes. When Isotropic factor is applied to T DSH (IsoDSH), it achieves an excellent capability of discriminating shape series in different sizes. Inspired by the histogram distribution of DSH algorithm, a novel SAH algorithm was defined based on SA algorithm. Its accuracy is greatly enhanced when compared to the SA algorithm. A quick anisotropic rescaled algorithm was proposed for the pre-process procedure of ISODSH method to improve the efficiency. This avoids the expensive computational consumption on the initial iterative method. Furthermore, The DSISAH algorithm was projected to intelligently integrate DSH, SAH and ISODSH methods, which broadens the describing perspectives of a model by embodying multiple feature descriptors. These three sub-algorithms complement each other in term of advantages and weakness, which results in a more balance method. From experiments, four key parameters of DSISAH algorithm were defined and verified. The optimal sample size, in terms of the number of description sample points for presenting a 3D shape under study, was achieved into a balance between the stability of the shape descriptors and the algorithm efficiency. The meaningful solid angle value range for the SAH shape descriptor was also identified, using the method of analyzing the effective range of the original histogram distribution. The maximal difference between two similar shapes was demarcated, according to the statistic of human awareness criterion for similarity evaluation. The optimized weighting solution that combines sub-algorithms together was also derived, via the method of approximation, to maximize its accuracy in similarity assessment. The feasibility of DSISAH algorithm is verified, by comparing its similarity evaluation and similar retrieval performance with DSH, SAH and IsoDSH methods. DSISAH algorithm shows a promising performance not only in shape categorization discrimination, but also in detail surface deformation and series identification. DSISAH algorithm was also implemented into a retrieval engine. A friendly user interface was developed for database management and shape query. With this search engine, users can conveniently retrieve the most similar shapes from the database or directly assess the similarity between the specified shapes using their favorite method among DSH, SAH, IsoDSH and DSISAH methods. This thesis solved a fundamental problem in design re-usage. The performance of the proposed new method is promising in a large range of products. The proposed DSISAH method could play a key role in model retrieval and act as an evaluation tool in customized design in virtually any product domains. It has the potential to leverage current prevailing search engine from currently text-based abstracted query to directly 3D shape-based search.