Abstract. Among those few hypotheses of Amazonian diversification amenable to falsification by phylogenetic and population genetics methods, three can be singled out because of their general application to vertebrates: the riverine barrier, the refuge, and the Miocene marine incursion hypotheses. I used phylogenetic and population genetics methods to reconstruct the diversification history of the upland (terra-firme) forest superspecies Xiphorhynchus spixii/elegans (Aves: Dendrocolaptidae) in Amazonia, and to evaluate predictions of the riverine barrier, refuge, and Miocene marine incursion hypotheses. Phylogeographic and population genetics analyses of the X. spixii/elegans superspecies indicated that the main prediction of the riverine barrier hypothesis (that sister lineages occur across major rivers) hold only for populations separated by ''clear-water'' rivers located on the Brazilian shield, in central and eastern Amazonia; in contrast, ''white-water'' rivers located in western Amazonia did not represent areas of primary divergence for populations of this superspecies. The main prediction derived from the refuge hypothesis (that populations of the X. spixii/elegans superspecies would show signs of past population bottlenecks and recent demographic expansions) was supported only for populations found in western Amazonia, where paleoecological data have failed to support past rainforest fragmentation and expansion of open vegetation types; conversely, populations from the eastern and central parts of Amazonia, where paleoecological data are consistent with an historical interplay between rainforest and open vegetation types, did not show population genetics attributes expected under the refuge hypothesis. Phylogeographic and population genetics data were consistent with the prediction made by the Miocene marine incursion hypothesis that populations of the X. spixii/elegans superspecies found on the Brazilian shield were older than populations from other parts of Amazonia. In contrast, the phylogeny obtained for lineages of this superspecies falsified the predicted monophyly of Brazilian shield populations, as postulated by the Miocene marine incursion hypothesis. In general, important predictions of both riverine barrier and Miocene marine incursion hypotheses were supported, indicating that they are not mutually exclusive; in fact, the data presented herein suggest that an interaction among geology, sea level changes, and hydrography created opportunities for cladogenesis in the X. spixii/elegans superspecies at different temporal and geographical scales.Key words. Aves, Amazonia, Miocene marine incursion hypothesis, phylogeography, population genetics, refuge hypothesis, riverine barrier hypothesis.Received March 10, 2003. Accepted February 10, 2004 The bird fauna of the Amazon lowlands in South America is the richest in the world, with over 1000 species, of which about 265 are endemic (Stotz et al. 1996;Nores 2000). As early as 1852, the British naturalist Alfred Russel Wallace put forward an evolutionary...