"Context-based courses" are increasingly used in an address to the major challenges that science education currently faces: lack of clear purpose, content overload, incoherent learning by students, lack of relevance to students, and lack of transfer of learning to new contexts. In this paper, four criteria for the design of context-based courses that would be successful in meeting these challenges are rehearsed. It is concluded that only a model based on "context as social circumstances" would meet the four criteria for success. From this, the notion of concept development is presented based on the idea of the production of coherent mental maps. The notion of transfer is discussed in terms of how such mental maps may be useful for understanding other contexts. The definitions of concept development and transfer give a clearer view of how exemplars of existing context-based approaches may be analysed to show their degree of facilitation of worthwhile science education. Research questions to be addressed in such analyses are presented. (Contains 3 figures.)