Tropical lake sediments are a significant source for the greenhouse gas methane. We studied function (pathway, rate) and structure (abundance, taxonomic composition) of the microbial communities (Bacteria, Archaea) leading to methane formation together with the main physicochemical characteristics in the sediments of four clear water, six white water and three black water lakes of the Amazon River system. Concentrations of sulfate and ferric iron, pH and delta C-13 of organic carbon were usually higher, while concentrations of carbon, nitrogen and rates of CH4 production were generally lower in white water versus clear water or black water sediments. Copy numbers of bacterial and especially archaeal ribosomal RNA genes also tended to be relatively lower in white water sediments. Hydrogenotrophic methanogenesis contributed 58+/-16% to total CH4 production in all systems. Network analysis identified six communities, of which four were comprised mostly of bacteria found in all sediment types, while two were mostly in clear water sediment. Terminal restriction fragment length polymorphism (T-RFLP) and pyrosequencing showed that the compositions of the communities differed between the different sediment systems, statistically related to the particular physicochemical conditions and to CH4 production rates. Among the archaea, clear water, white water, and black water sediments contained relatively more Methanomicrobiales, Methanosarcinaceae and Methanocellales, respectively, while Methanosaetaceae were common in all systems. Proteobacteria, Deltaproteobacteria (Myxococcales, Syntrophobacterales, sulfate reducers) in particular, Acidobacteria and Firmicutes were the most abundant bacterial phyla in all sediment systems. Among the other important bacterial phyla, clear water sediments contained relatively more Alphaproteobacteria and Planctomycetes, whereas white water sediments contained relatively more Betaproteobacteria, Firmicutes, Actinobacteria, and Chloroflexi than the respective other sediment systems. The data showed communities of bacteria common to all sediment types, but also revealed microbial groups that were significantly different between the sediment types, which also differed in physicochemical conditions. Our study showed that function of the microbial communities may be understood on the basis of their structures, which in turn are determined by environmental heterogeneity.