Results/Conclusions: Jaccard similarity indexes for recruited species indicated that primary and secondary forest shared > 50 % of the species, while pastures excluded from cattle since June 2006, shared very few species with secondary forest (0.17) and shared even less species with primary forest (0.02). Only one late-successional species (Ficus perforata) and four pioneers (Trichospermum galeottii, Heliocarpus appendiculatus, Eupatorium galeottii and Cecropia obtusifolia) were present in the seed rain and the recruited community at all three habitats. In primary and secondary forest, these five species had very low limitation in dispersal (0-3%), but extremely high limitation in establishment (83-100%). In pastures, all species showed very high dispersal limitation (86-98%), except for two wind-dispersed species, Eupatorium (60%) and Heliocarpus (38%) both of which showed high establishment limitation (79 and 86%). Cecropia, regardless of its immense dispersal limitation in pastures (97%), had the lowest establishment limitation (67 %) comparing to the other four species. Overall species and habitats, there was no a correlation of seed density on recruits density (r = 0.14, P > 0.1, N = 24). For primary forest there was a negative correlation of seed density on recruits density (r = -0.48, P < 0.05, N = 14). Mostly all species showed an extremely high dispersal and establishment limitation in pastures. However, limitation in establishment may not always be explained by the lack of seed arriving at a given year.  Natural succession in pastures is very slow and the resulting secondary forest will show a completely different plant composition compared to the primary forest.