The distances that individuals disperse, from their natal site to the site of first breeding and between breeding sites, have important consequences for the dynamics and genetic structure of a population. Nearly all previous studies on dispersal have the problem that, because the study area encompassed only a part of the population, emigration may have been confounded with mortality. As a result long-distance dispersers may have been overlooked and dispersal data biased towards short distances. By studying a virtually closed population of Seychelles warblers Acrocephalus sechellensis we obtained almost unbiased results on several aspects of dispersal. As in the majority of other avian species, natal dispersal distance was female biased in the Seychelles warbler. Female offspring also forayed further from the natal territory in search of breeding vacancies than male offspring. The sex bias in natal dispersal distance did, however, depend on local breeding density. In males, dispersal distance decreased as the number of territories bordering the natal territory increased, while in females, dispersal distance did not vary with local density. Dispersal by breeders was rare and, unlike in most species, distances did not differ between the sexes. We argue that our results favour the idea that the sex bias in natal dispersal distance in the Seychelles warbler is due to inbreeding avoidance and not resource competition or intrasexual competition for mates.