This paper considers the use of secondary data analysis in educational research. It addresses some of the promises and potential pitfalls that influence its use and explores a possible role for the secondary analysis of numeric data in the "new" political arithmetic tradition of social research. Secondary data analysis is a relatively under-used technique in Education and in the social sciences more widely, and it is an approach that is not without its critics. Here we consider two main objections to the use of secondary data: that it is full of errors and that because of the socially constructed nature of social data, simply reducing it to a numeric form cannot fully encapsulate its complexity. However, secondary data also offers numerous methodological, theoretical and pedagogical benefits. Indeed by treating secondary data analysis with appropriate scepticism and respect for its limitations, by demanding that tacit assumptions about the unreliability of secondary data are applied equally to other research methods, and crucially by combining secondary data analysis with small-scale in-depth work, this paper argues for a return to prominence of secondary data analysis in its own right as well as becoming a central component of the new political arithmetic tradition of social research.