Advocates of constructivist science recommend that school science begins with children’s own constructions of reality. This notion of the way in which students’ knowledge of science grows is closely paralleled by recent research on teachers’ knowledge. This paper draws on case study evidence of teachers’ work to show how two experienced teachers’ attempts to develop alternative ways of teaching science involved reframing their previous patterns of understanding and practice. Two alternative interpretations of the case study evidence are offered. One interpretation, which focuses on identifying gaps in the teachers’ knowledge of science teaching, leads to theconstructivist paradox. The second interpretation explores theconstructivist parallel, an approach which treats the process of teachers’ knowledge growth with the same respect as constructivists treat students’ learning of science. This approach, the authors argue, is not only more epistemologically consistent but also opens up the possibilities of helping teachers lead students towards a constructivist school science.