In this paper we investigate the validity of two education requirement measures. This is important because a key part of the ongoing discussion concerning overeducation is about measurement. Thanks to the Dutch Institute for Labour Studies, we have been given a unique opportunity to compare two education requirement measures: first, Huijgen's rather dated measure and, second, the new measure of Statistics Netherlands. To investigate the validity of these two measures we apply two existing methods as well as develop and apply a new method. This new method can also be used when only one measure is available. In the end, the Statistics Netherlands measure proved to be more valid than the Huijgen measure. (c) 2005 Elsevier Ltd. All rights reserved.