In the past decades, much work has been done in science education on large–scale curriculum development, ranging from a “structure–of–the–discipline” approach to STS. At the same time, research on students' ideas has drawn attention to the underestimated problems of learning and teaching, which may largely explain the limited success of the curriculum efforts as far as cognitive learning is concerned. Proposed solutions are mainly inspired by a constructivist cognitive science perspective and are formulated as general teaching strategies that aim at a more or less forced process of “conceptual change.” However, in our view, “developmental research” is needed in which small–scale curriculum development is cyclically coupled to indepth classroom research of teaching–learning processes. Such research should resuit in worked out examples of successful ways of teaching, according to new conceptual curriculum structures. Designing such “didactical” structures constitutes a longer term research program, which asks for international exchange and cooperation. © 1995 John Wiley & Sons, Inc.