The Kibale Forest, western Uganda, is the only site where studies have compared the impact of elephants on rainforest regeneration in logged and unlogged control areas. Elephants used heavily logged areas more than lightly logged and unlogged areas. Forest gaps were used more by elephants than closed-canopy areas and large gaps more than small ones. Gaps were larger in logged than unlogged forest. There were lower densities of young trees (saplings and poles) and a higher incidence of elephant damage to them in heavily logged forest than in lightly logged and unlogged sites. Elephant use of an area and damage to young trees was inversely or unrelated to the density of young trees and directly related to the density of herbaceous tangle. Heavy logging resulted in large areas of herbaceous tangle, which attracted elephants who suppressed forest regeneration by damaging young trees and perpetuating the herbaceous tangle. The tangle directly competed with regeneration of young trees while also attracting elephants and rodents (seed and seedling predators) and facilitating increased windthrow of trees. Selective browsing of young trees by elephants affected rates of regeneration, growth form and species composition. Rather than remove elephants, a more effective and humane approach to long-term management of logging is to reduce logging offtake and incidental damage caused by timber extraction.