The considerable variability found in post-dispersal seed predation and the absence of consistent directional trends (e.g., with reference to seed size) has made it difficult to predict accurately the impact of seed predators on plant communities. We examined the variation attributable to location, seed density and seed burial on the removal of seeds of three tree species: Fraxinus excelsior, Taxus baccata and Ulmus glabra. Experiments were undertaken in five deciduous woodlands in Durham, U.K., and the relative importance of vertebrate and invertebrate seed predators was assessed using selective exclosures. In all five woodlands, seed removal was greatest from treatments to which vertebrates had access, and losses attributable to invertebrates were negligible. Rodents, in particular Apodemus sylvaticus (Muridae) and Clethrionomys glareolus (Cricetidae), were the principal seed consumers in these woodlands. Unidentified vertebrate seed predators (probably birds, rabbits and/or squirrels) appeared to be significant seed removers in three of the five woodlands. Rates of removal differed among the three tree species, increasing in the following order Fraxinus < Taxus < Ulmus but were not related to seed mass. The major effect influencing rates of seed removal was seed burial, which halved rates of seed removal overall. The effect of seed burial was a function of seed size. The larger seeds of Taxus realising little benefit from seed burial whereas encounter of the smaller Ulmus seeds fell by almost two-thirds. Removal was density-dependent for all three species. However, the relative increase in seed encounter through an increase in seed density was a negative function of seed size. This suggests that, for large seeds, the opportunity to escape seed predation via burial or reduced seed density is limited. These results reveal a number of parallels with other studies of post-dispersal predation and identify several generalities regarding the interaction between plants and post-dispersal seed predators. Comparison of the seed predation results with actual seedling distributions suggests that seed predators may influence regeneration of Ulmus glabra but probably play a lesser role in the dynamics of Taxus baccata and Fraxinus excelsior.