The abundance of forest tree species may be locally limited by the inability of species to disperse to all sites suitable for germination and establishment. This phenomenon of “dispersal limitation” has been suggested to promote the maintenance of high species diversity in tropical forests by slowing down competitive exclusion. We present the first direct experimental evaluation of dispersal limitation in tropical forests, and of its importance relative to other factors affecting recruitment. Seed addition, litter removal and light availability effects on seedling establishment were evaluated in a full factorial experiment on two African mahogany species, Entandrophragma utile and Khaya anthotheca. Seedling recruitment after 18 months, was ∼22 times greater in seed addition treatments (with addition rates corresponding to the tail of seed shadows of large reproductive trees) compared to controls. This seed addition effect was an order of magnitude greater than either gap or litter removal effects, with similar results observed in both logged and primary forest locations. We conclude that dispersal limitation strongly limits seedling establishment for the two species studied. An important “applied corollary” of this result is that seed supplementation may substantially increase local abundance of valuable or rare species in disturbed or managed tropical forests.