Amazonia is an environmentally heterogeneous and biologically megadiverse region, and its biodiversity varies considerably over space. However, existing knowledge on Amazonian biodiversity and its environmental determinants stems almost exclusively from studies of macroscopic above‐ground organisms, notably vertebrates and trees. In contrast, diversity patterns of most other organisms remain elusive, although some of them, for instance microorganisms, constitute the overwhelming majority of taxa in any given location, both in terms of diversity and abundance. Here, we use DNA metabarcoding to estimate prokaryote and eukaryote diversity in environmental soil and litter samples from 39 survey plots in a longitudinal transect across Brazilian Amazonia using 16S and 18S gene sequences, respectively. We characterize richness and community composition based on operational taxonomic units (OTUs) and test their correlation with longitude and habitat. We find that prokaryote and eukaryote OTU richness and community composition differ significantly among localities and habitats, and that prokaryotes are more strongly structured by locality and habitat type than eukaryotes. Our results 1) provide a first large‐scale mapping of Amazonian soil biodiversity, suggesting that OTU richness patterns might follow substantially different patterns from those observed for macro‐organisms; and 2) indicate that locality and habitat factors interact in determining OTU richness patterns and community composition. This study shows the potential of DNA metabarcoding in unveiling Amazonia's outstanding diversity, despite the lack of complete reference sequence databases for the organisms sequenced.