Abstract. The increasing importance of better customisation of industrial products has led to development of configurable products. They allow companies to provide product families with a large number, typically millions of variants. Description and management of a large product variety within a single data model is challenging and requires a solid conceptual basis. This management differs from the schema evolution of traditional databases and the crucial distinction is the role of schema. For configurable products, the schema changes more frequently and more radically. In this paper, the characteristics that distinguish configurable products from traditional data modelling and management are investigated taking into account both the evolution of the schema and the instances. In this respect, the existing data modelling approaches and product data management systems are inadequate. Therefore, a new conceptual framework for product data management systems of configurable products that a...