Over recent years the findings of a number of quantitative research studies have been published in the UK on gender and achievement. Much of this work has emanated from Stephen Gorard and his colleagues and has not only been highly critical of existing approaches to handling quantitative data but has also suggested a number of alternative and, what they claim to be, more valid ways of measuring differential patterns of achievement and underachievement between groups. This article shows how much of this work has been based upon rather under-developed measures of achievement and underachievement that tend, in turn, to generate a number of misleading findings that have questionable implications for practice. It will be argued that this body of work provides a useful case study in the problems of quantitative research that fails to engage adequately with the substantive theoretical and empirical literature and considers some of the implications of this for future research in this area. (Contains 1 figure and 1 table.)