The unified neutral theory of biodiversity and biogeography provides a dynamic null hypothesis for the assembly of natural communities. It is also useful for understanding the influence of speciation, extinction, dispersal and ecological drift on patterns of relative species abundance, species-area relationships and phylogeny. Clark and McLachlan argue that neutral drift is inconsistent with the palaeorecord of stability in fossil pollen assemblages of the Holocene forests of southern Canada. We show here that their analysis is based on a partial misunderstanding of neutral theory and that their data alone cannot unambiguously test its validity.