We compared the seed fate of two animal-dispersed, large-seeded timber species (Dipteryx panamensis [Fabaceae] and Carapa guianensis [Meliaceae]) in logged and fragmented forests with that for continuous forest in northeastern Costa Rica. For both species, we quantified rates of seed removal (an index of vertebrate predation) and the fate of dispersed seeds (those carried away from their original location that either germinated or were not subsequently removed within three months). We predicted that (1) fewer seeds would be dispersed by vertebrates in fragmented forest than in continuous forest due to low population abundances after hunting and/or loss of suitable habitat, and (2) seed predation rates would be higher in forest fragments than in continuous forest due to high abundance of small-bodied seed consumers. We compared three forest fragments currently managed for timber (140-350 ha) and a large reserve of continuous forest (La Selva, 1500 ha and connected to a national park). An exclusion experiment was performed (seeds placed in the open vs. seeds within semipermeable wire cages; 5 cm mesh size) to evaluate the relative roles of large and small animals on seed removal. Seed germination capacity did not differ among all four sites for both species. Removal of Dipteryx seeds was higher in forest fragments (50% removal within 10 days and related to the activity of small rodents) compared to La Selva (50% removal after 50 days). Also, more Dipteryx seeds were dispersed at La Selva than in fragmented forests. Contrary to our predictions, removal of Carapa seeds was equally high among all four sites, and there was a trend for more seeds of Carapa to be dispersed in fragments than in La Selva. Our results suggest that fragmentation effects on tree seed fate may be specific to species in question and contingent on the animal biota involved, and that management strategies for timber production based on regeneration from seed may differ between forest patches and extensive forests.