The literature has typically assumed that the effect of overeducation on wages is constant over the conditional wage distribution. In this paper, we use quantile regression and data from 12 European countries to show that the overeducation wage effect may differ largely across segments of the distribution. By differentiating between quantiles, we discriminate between groups of workers with different (unobservable) skills. We find that the detrimental effects of overeducation among the high-skilled are indeed higher than among the lowskilled. This finding lends support to the view that overeducation is an event that reduces the worker’s potential productivity, regardless of his skills.