Since the 1970s there has been considerable debate among sociologists of education about the macro-micro gap in educational analyses. However, educational research remains divided largely into the study of large-scale phenomena such as social systems and national policies on the one hand, and case-studies of individual schools and social interaction on the other. This split has had a number of unfortunate consequences for the development of the field. Most importantly, the dominant conceptions of structure and agency employed in the sociology of education are characterised by a dualism which makes it difficult to conceptualise adequately the processes involved in social change. In this paper, I briefly describe this structure-agency dualism before critically examining three attempts which have been made to address this problem. The ability of structuration theory to overcome this dualism is then examined, and I conclude by arguing that this approach offers an important new direction for the sociology of education.