The Plateado or Dorado - Brachyplatystoma rousseauxii (Pimelodidae, Siluriformes) is a commercial migratory catfish species with one of the most surprising and enigmatic life histories in the Amazon basin, involving the largest migration known for a freshwater species, between the estuary and the head waters in the Andean piedmont. The aim of the present work was to determine the molecular phylogenetic position of the Plateado in the Pimelodidae family and its population genetic structure in the Upper Madera (Villa Bella – VB, Cachuella Esperanza – CE, Puerto Maldonado – PM, Rurrenabaque – RU, Puerto Villarroel - PV) and Western Amazon (Iquitos - IQ) basins (Bolivia and Peru). The phylogenetic relationships were defined through a Maximum Likelihood (ML) analysis of nucleotide sequences of two mitochondrial (Control Region – CR, ~ 900 pb, 32 taxa; Cytochrome Oxidase 1 – CO1, ~ 650 bp, 61 taxa), and a nuclear fragment (F-reticulon4 - RTN4, ~1700 bp, 38 taxa). The population genetic structure was evaluated through the length polymorphism of nine microsatellites (284 inds) and CR sequence variations (461 inds + 45 from Brazil available in GenBank). Microsatellites frequencies variations were used to identify through a Bayesian approach (BAPS) the most probable panmictic units (clusters) in the whole data, after previous demonstration of a deviation to Hardy-Weinberg Equilibrium (HWE). The ML phylogenetic concatenated analysis showed the Pimelodidae family as a monophyletic group, with the genera Phractocephalus and Leiarius as basal lineages. The most notable results in the phylogeny were the not well-supported monophyly (77%) of the tribe Brachyplatystomatini and the non-monophyly of Brachyplatystoma. Only the morfologically defined subgenus Malacobagrus (B. rousseauxii + (B. filamentosum + B. capapretum)) was recovered as monophyletic. These results suggest that Brachyplatystoma could contain Platynematichthys or be restricted to the subgenus Malacobagrus, and the other species be related to distinct (earliest) genera, in agreement with another study carried out in parallel with other markers.Microsatellite analysis of the whole data (Western Amazon + Upper Madera) showed a significant departure of the HWE expectations, as well as the analysis of the whole data from the Upper Madera region. In the light of these results, the Bayesian approach has been implemented, showing that at least three clusters (1, 2, 3) are present in the Upper Madera and Western Amazon basins with partial overlapping distribution.To the margin of the cluster identification, it was evident the significant difference between Western Amazon (Iquitos region) and the Upper Madera basin.The genealogical analysis (ML) of the CR sequences showed a generalized comb-like topology without group of haplotypes with common ancestry. On the other hand, CR frequency analysis showed the conformation of four haplogroups associated to geography. One haplogroup was identified along the main axis of the Amazonas-Solimões, from Belem (Brazil) to Iquitos (Peru), and three other haplogroups were observed in the Upper Madera basin (VB; CE+PM; RU+PV), positioned in a downstream - upstream pattern.Hence, we observed on the one hand three genetic populations (clusters), distributed in partially overlapping geographical areas, and on the other hand four haplogroups, positioned according to a geographical pattern. The most probable scenario involves a homing behavior of individuals from cluster 1 (homing at the scale of large watersheds), which prefer or tend to return to the Madera basin, with the three populations coexisting within the upper Madera because they reproduce at different moments (phenology) or different places (spatial segregation). Finally, the results are discussed in the light of previous results in the Amazon basin and the threats to the species in the Madera basin (p.e. fragmentation by dams, overfishing, climate variability, among other).