The study stresses that it is made in the context of enterprise retrieval systems. The findings and the issues may, however, also be relevant for traditional bibliographic databases using thesauri or other kinds of controlled vocabularies. It is, thus, assumed that the issues discussed by both L&E and the present article may have very broad implications. We found the findings surprising. In general, the expanded search queries performed better than metadata searches on both recall and precision. The metadata searches resulted in the lowest precision, which is remarkable because the human indexers were tailored specifically to meet information tasks such as those search tasks constructed for the experiment. It is also remarkable that the expanded search retrieved more of the highly relevant documents (which is described as discouraging, Lykke & Eslau, p. 93).