Summary.The investigation reported in this paper studied the relationship between both GCE O‐level grades and teachers' estimates of A‐level grades, and the actual A‐level grades obtained, in individual subjects, by a sample of applicants for university places. Moderate levels of correlation were reported in both cases, although the teachers' estimates appeared to be slightly better predictors of the A‐level grades. There also appeared, in both cases, to be some important differences between the levels of correlation reported in individual subjects. In addition it was observed that the teachers, on average, tended to over‐predict the A‐level grades. It was noticeable that both the O‐level grades and the teachers' estimates of the A‐level grades were bunched together in a very narrow range at the top of the grading scale. The effect of this restriction amongst the predictor variables is discussed in terms of its influence on the levels of correlations reported. The application of certain corrections that are intended to compensate for this effect is