Design of sheet metal forming tooling is currently based on that experienced tooling designers with good knowledge of how stamping tools previously have been designed and operated in production, apply their knowledge when making a new design. For retrieving former designs, they often need to rely on their good memory. In this paper, an automatic method for retrieving relevant former cases is presented. A major challenge is defining the similarity between the current and the former cases i.e., finding the relevant parameters to include in the CBR (Case-based reasoning) search. This is here addressed by using CAD model parameters both from former components and the tooling for their production. By interviewing tooling designers in industry, a set of relevant parameters has been identified. To arrive at the correct weight of each parameter, a genetic algorithm has been used to optimize the search results. This resulted in a quick and automated way of retrieving the most relevant former cases and presenting them to the designer. The method has been tested on actual cases with promising results. This has the potential of making sheet metal part and tooling design less reliant on memory and experience.