This paper presents a study on die-casting die design based on case-based reasoning (CBR). Being different from a traditional die-casting CAD system, this system stores the knowledge and experience as cases, and modifies a previous case to satisfy a new situation. By saving the new solution in the case library after the latest problem has been solved, the system is able to perform self-learning from the experience accumulated. The work includes: 1. Developing a case-based reasoning environment for diecasting die design, so that the system can self-learn from the previous consultation itself and become more efficient. 2. Developing an effective case representation and adaptation methodology for reusing previous design resources efficiently. This paper focuses mainly on the design of die-casting dies. However, the proposed framework is generally adaptable for other processes (such as fixture design and injection mould design) by customising specific process features and encoding process design rules.