Edge creation has a pronounced influence on the understory vegetation, but the effects of edges on seedling species recruitment are still poorly understood. In Central Amazonia, 9–19 years after fragmentation, we recorded species richness and net seedling recruitment rate in 1 ha blocks exposed to none, one, or multiple edges within forest fragments. One-hectare blocks were located in the center (no edge), the edge (one edge), the corners (two edges) of 10 and 100 ha fragments, and in a 1 ha fragment (four edges). In 1991, we counted all tree seedlings 5–100 cm tall found within permanent 1 m² plots located within the 1 ha blocks. In May 1993, we manually removed all seedlings that were smaller than 1 m tall from the permanent plots. Six years and five months later (October 1999), all new seedlings recruited into the plots were counted and classified into distinct morphospecies. Species richness of recruited seedlings, scaled by total seedling density, declined from the center to the edge, the corner blocks, and then to the 1 ha fragment. Overall, the four-edged, 1 ha fragment had the poorest species richness and the non-edged 100 ha central block the highest. The total number of recruited individuals was 40 percent less than that previously present, with the 100 ha corner having the lowest recruitment. Pairwise comparisons showed that species similarity was related to edge number for the 100 and 1 ha fragments. Species rank/abundance curves showed that a subset of species was common in all blocks within the fragments, and that the 100 ha center held more rare species than any other 1 ha block. This study demonstrated that, in a given fragment patch, the number of tree seedling species recruited varied inversely with the number of edges.