Traditional techniques of 3D data retrieval using names and subjective attributes are not robust and are difficult to automate over large 3D repositories. Certain techniques developed for search and classification of 2D engineering designs, are in general, difficult to extend to 3D models. These issues are addressed by a system for automated search and classification for 3D CAx models based on their geometric "shape," which is often an indication of design, analysis and manufacturing process similarity. A new method and system are provided for representing 3D shape as a composition of multiple 2D image projections, which are transformed using the Discrete Fourier and Harr Wavelet transforms. Key coefficients of the transforms are then stored in the 3D model repository and are used to efficiently search and classify such repositories.