The present research aims to more fully explore the issues of performance differences in higher education assessment, particularly in the context of a common measure taken to address them. The rationale for the study is that, while performance differences in written examinations are relatively well researched, few studies have examined the efficacy of anonymous marking in reducing these performance differences, particularly in modern student populations. By examining a large archive (N = 30674) of assessment data spanning a twelve-year period, the relationship between assessment marks and factors such as ethnic group, gender and socio-environmental background was investigated. In particular, analysis focused on the impact that the implementation of anonymous marking for assessment of written examinations and coursework has had on the magnitude of mean score differences between demographic groups of students. While group differences were found to be pervasive in higher education assessment, these differences were observed to be relatively small in practical terms. Further, it appears that the introduction of anonymous marking has had a negligible effect in reducing them. The implications of these results are discussed, focusing on two issues, firstly a defence of examinations as a fair and legitimate form of assessment in Higher Education, and, secondly, a call for the re-examination of the efficacy of anonymous marking in reducing group performance differences.