Leaf-cutting ant nests represent a potential source of disturbance within Neotropical forests that might favour selective establishment of some plant species. To test this hypothesis, I counted the number and determined the composition of plant species and individuals established in 10 abandoned Atta cephalotes nests and adjacent understory plots in an old-growth forest of Costa Rica. Specifically, I evaluated whether abandoned leaf-cutting ant nests differentially affect plant assemblages according to their 1) regeneration status, 2) seed size, and 3) dispersal mode. No differences were found in the relative abundance of species (and individuals) with different regeneration status, seed size, and dispersal mode between understory plots and abandoned Atta nests. Moreover, entire sites (abandoned ant nests and their nearby understory plots) were frequently grouped together according to those characteristics, suggesting that local effects were stronger than the effect of nests. Four non-mutually exclusive factors are discussed to explain the lack of a consistent effect of abandoned ant nests on plant assemblage composition: refuse dump location, spatial heterogeneity, dispersal limitation, and the local species pool. Chance events in the dynamics of both ant and plant species populations apparently restrict the effect of abandoned nests to an ecological time. Abandoned ant nests, however, might favour selective establishment of some plant assemblages if their refuse piles are located on the soil surface and if the potentially favoured plant species are locally available.