Habitat models are a common tool to assess the factors driving habitat selection, and are a pre-requisite for many conservation measures. I investigated habitat selection of the white-backed woodpecker (Dendrocopos leucotos), a species highly dependent on dead wood and regarded as typical for primeval forests, in Vorarlberg (Western Austria), Eastern Switzerland and Liech-tenstein. Here, the species has increasingly been found in managed forests. I used site-occupancy models to compare a priori expectations as well as models built a poste-riori. Occupancy probability was mainly related to variables concerning forest structure: it was positively related to the average dbh of live trees and snags, the mean diameter of lying dead wood, and the availability of saproxylic beetles. In addition, it was negatively related to elevation above sea level. In general, forest structure seemed to be more important than topography. Analyses at different spatial scales indicated that small patches containing habitat factors asso-ciated with high occupancy probability are sufficient for the occurence of white-backed wood-peckers in the study area. Study plots with high occupancy probability were distributed evenly over the whole study area. Altogether, forest management in white-backed woodpecker habitats is possible, but should be done extensively. Patches with old, deciduous forests and high quantities of large dead wood should be retained or created to enhance habitat quality for this species.