There exist numerous design rationale systems that convert captured information into structured design rationale while providing rationale representation and retrieval. These systems woefully neglect the design rationale that is present in legacy CAD such as 2D drawings and 3D models. The dissertation addresses the issues that arise when dealing with the capture, representation and retrieval of design rationale from the 2D legacy CAD data, specifically the non-form related data (e.g. text and symbols). A definition for design rationale in the CAD domain is presented which forms the basis of the proposed approach. The approach uses a unique context-aware inference system to capture design rationale from legacy CAD data. A brief explanation of context is provided along with the advantages of using context for this task. The need and use of an inference system is detailed. Additionally a prototype system is implemented to address these issues from a software system point of view. A verification process is suggested that will validate the design rationale captured by the system to that captured by human re-designers.