Résumé : Labour economists typically assume that pay differences between occupations can be explained with variations in productivity. The empirical evidence on the validity of this assumption is surprisingly thin and subject to various potential biases. The authors use matched employer-employee panel data from Belgium for the years 1999-2006 to examine occupational productivity-wage gaps. They find that occupations play distinct roles for remuneration and productivity: while the estimations indicate a significant upward-sloping occupational wage-profile, the hypothesis of a flat productivity-profile cannot be rejected. The corresponding pattern of over- and underpayment stands up to a series of robustness tests.