An analytical model of competitive coexistence in spatial habitats, modified to address habitat destruction, predicts that the most abundant species can be among the first species driven extinct by habitat destruction, given that abundant species are the poorest dispersers and best competitors. This contrasts with the classical view of biased extinction of rare species. Here we explore the robustness of this prediction both analytically and in spatially explicit simulations of more realistic cases. The prediction proved surprisingly robust. The poorest dispersers, which in this model generally are the best competitors and may be the most abundant species, were among the first driven extinct by habitat destruction whether they were abundant or rare, had short or long range dispersal, or reproduced continuously or periodically; whether competitive displacement was immediate or gradual; whether habitat destruction was clumped, uniform, or random and whether destruction occurred at once or progressively; and whether the habitat was large or small. The amount of destruction sufficient to produce extinctions changed considerably as model assumptions changed, but the biased extinction remained. The underlying reason for the robustness of our conclusions is the broad assumption that inferior competitors persist by virtue of greater dispersal ability and/or lower mortality rates. Further work on the forces allowing multispecies coexistence is thus essential for understanding the effects of habitat destruction on extinction.