This study investigates the species-area relationship in Amazonian forest fragments. The relationship, as detected by pitfall trapping of with exponent z =0.17. Several independent samples show this same power function with similar z values. Furthermore, the sampled community shows a lognormal distribution of "activity densities", though this does not necessarily mean that the underlying distribution of species-area curve was achieved by the removal of generalist species from consideration. These species invade from clear-cut areas, penetrating small fragments, but not reaching the interior of large fragments. Some analyses of trapping efficiency were also performad. The effects of trap size and trap spacing on the cath were investigated, showing that large and small traps provide sowewhat different community samples. Similar, traps that are too close together show a depressed catch of large-bodied individuals. Lastly, some explanations for the observed species-area relationship are considered, along with implications for the design of nature reserves.