National assessment has been introduced in many countries to monitor children's progress in mathematics and to exert some influence over their curriculum experience. In England, children undertake national tests whose results are also used to compare schools. Reflecting thinking in mathematics education, these tests have tended to embed mathematical tasks in supposedly "realistic" settings. Such "realistic" contexts might be expected to have a variety of benefits. Perhaps working class children will benefit from the "relevant" nature of such contexts? Perhaps girls will perform better on such items? In order to address such questions, the authors have undertaken quantitative and qualitative research with 10-11 and 13-14 year-old children to explore their interpretation of and performance on English national mathematics test items, with a special focus on the validity and fairness of "realistic" items. This text draws on the work of Bernstein and Bordieu to make sense of findings which suggest that the validity of such items may vary by social class and sex.