Construction of hydroelectric dams in tropical regions has been contributing significantly to forest fragmentation. Alterations at edges of forest fragments impact plant communities that suffer increases in tree damage and dead, and decreases in seedling recruitment. This study aimed to test the core-area model in a fragmented landscape caused by construction of a hydroelectric power plant in the Brazilian Amazon. We studied variations in forest structure between the margin and interiors of 17 islands of 8-100 hectares in the Tucuruí dam reservoir, in two plots (30 and >100m from the margin) per island. Mean tree density, basal area, seedling density and forest cover did not significantly differ between marginal and interior island plots. Also, no significant differences were found in liana density, dead tree or damage for margin and interior plots. The peculiar topographic conditions associated with the matrix habitat and shapes of the island seem to extend edge effects to the islands' centers independently of the island size, giving the interior similar physical microclimatic conditions as at the edges. We propose a protocol for assessing the ecological impacts of edge effects in fragments of natural habitat surrounded by induced (artificial) edges. The protocol involves three steps: (1) identification of focal taxa of particular conservation or management interest, (2) measurement of an "edge function" that describes the response of these taxa to induced edges, and (3) use of a "Core-Area Model" to extrapolate edge function parameters to existing or novel situations.