Abstract 1 Most plant-feeding insects show some degree of specialization and use a variety of cues to locate their host. Two main mechanisms of host location, primary attraction and random landing, have been investigated for such insects. 2 Research has led to contradictory conclusions about those hypotheses, especially for wood-feeding insects; however, recent studies suggest that both mechanisms may take place in a single taxon but at different scales. 3 We developed a field experiment to test the hypothesis that primary attraction occurs at larger scale and random landing at finer scale in wood-feeding insects. Landing rates, measured using sticky traps, were compared first between patches and then between individual trees according to their distance to a baited central tree. 4 Polynomial functions describing landing rate to distance relationships were compared with a function produced by a null model describing what should occur under the random landing hypothesis. Scolytidae and Cerambycidae (Coleoptera) responded to volatiles at the patch scale, supporting the primary attraction hypothesis, but the landing patterns of some groups at finer scale matched closely the predictions of our null model, giving support to the random landing hypothesis. 5 Our results show that the primary attraction and random landing hypotheses are not mutually exclusive and that prelanding use of host-produced volatile is scale-dependant. Scale considerations should thus be included in the study of prelanding host-selection of wood-feeding insects.