Much grading of student work is based, overtly or tacitly, on assumptions derived from scientific measurement. However, the practice of grading and the cumulation of grades into an overall index of achievement are socially constructed activities that fall a long way short of what is expected of scientific measurement. If scientific measurement is an unattainable ideal, how should the summative assessment of student achievement be approached? The case is argued that the professional judgement of assessors has to be given prominence, and that this implies a sustained commitment to developmental work at institutional and sectoral levels. Some suggestions along these lines are outlined.