Small forested islands isolated from surrounding tropical forest by new reservoirs provide a norm for interpreting the effects of fragmentating the forest by intervening agriculture and pasture, and an effective means of investigating the ecological organization of the mainland forest. Water is a more effective barrier to immigration and a more neutral matrix than field or pasture. On forest fragments isolated by water, the effects of fragmentation are minimally confounded with effects of the matrix, while forest fragments surrounded by fields and pastures may suffer intrusions from fire, domestic animals, and other pests. On small islands, some species will go extinct. If the extinction of a species is followed by increase in its prey or competitors, we may provisionally assume that on the mainland, these prey and competitors are limited by the species now absent from the island. If, when a tree's seed disperser disappears, new seedlings of that tree no longer appear, that tree's regeneration presumably depends on its seed disperser. Islands in new reservoirs are the tropical forest ecologist's closest analogue to the exclusion experiments so effective in understanding the ecology of inter-tidal communities. Small islands in reservoirs can also serve as systems of replicates for experimental analysis of the causes of regulation of selected populations. We review work on small islands isolated in 1914 by Panama's Gatun Lake, islands isolated in 1986 by Venezuela's Lake Guri, and islets isolated in 1994 at Saint-Eugene in French Guiana. The more recently the islands have been isolated, the more can be learned from them. The Saint-Eugene Fragmentation Project is particularly important because it is only one of the three in true rainforest and studies there have been done before and after fragmentation.