This study addresses the question of how to represent students' thinking about a complex topic in a way that is most useful to guide curriculum development, instructional decisions, and assessment of student learning. The development of a coding scheme that represents students' ideas along two dimensions, compatibility with expert propositions and elaboration of ideas, is described and applied to the analysis of interviews with four pairs of urban fifth-and sixth-grade students. Results indicate that decomposition is a linchpin concept for building an understanding of nutrient cycling, and that ideas which relate to children's natural framework theories provide a generative foundation for further learning. Finally, it is suggested that the elaboration dimension of the bidimensional coding scheme provides an indication of students' possession of valuable scientific habits of mind.The constructivist perspective that is influencing K-12 science education nationally (e.g., see American Association for the Advancement of Science [AAAS], 1993; National Research Council [NRC], 1996) holds that knowledge acquisition is not the mere accumulation of new facts, but a process that entails restructuring of knowledge as learners make sense of new information (Smith, Carey, & Wiser, 1985; von Glasersfeld, 1993). The crucial question for curriculum developers and science teachers is what degree of restructuring typically is necessary for different science concepts for children of different ages. Whether children's ideas need to undergo weak or strong restructuring (Carey, 1985), also seen as conceptual development, resolution, or exchange (Pines & West, 1986), has consequences for instruction. Children may experience different degrees of emotions and resistance to learning depending on the type of restructuring they encounter (Barker & Can, 1989b). Also, teachers need to choose whether to put time and effort into using a conceptual change teaching method (Posner, Strike, Hewson, & Gertzog, 1982), which may not be necessary or productive when strong restructuring is not needed (Barker & Carr, 1989b;Muthukrishna, Camine, Grossen, & Miller, 1993).The language and diagrams used to represent students' ideas about science phenomena can influence how instruction aimed at facilitating restructuring in those domains is shaped. For instance, describing students' ideas as "misconceptions" can imply that they are barriers to instruction and therefore require radical restructuring, whereas the terms "prior conceptions" or "alternative conceptions" imply that children and scientists have alternative views that can be negotiated through an evolutionary, rather than revolutionary approach to fostering cognitive growth (Abimbola, 1988, citing West, 1982. Other ways of describing children's ideas that can influence instructional decisions are in terms of how compatible their commonsense framework theories are with scientific theories (Vosniadou, 1994;Wellman & Gelman, 1992), and in terms of the match between the ontologic categories int...