It was recently proposed that the genus Bacteroides should be restricted to Bacteroides fragilis (the type species) and closely related organisms (viz., B. caccae, B. distasonis, B. eggerthii, B. merdae, B. ovatus, B. stercoris, B. thetaiotaomicron, B. uniformis, and B. vulgatus). By contrast, the moderately saccharolytic, predominantly oral Bacteroides species, which include B. melaninogenicus, B. oralis, and related species, form a phenotypically and phylogenetically coherent group of species which differ so significantly from the emended description of the genus Bacteroides that they should not be classified in the same genus. Therefore, we formally propose that these species be reclassified in a new genus, Prevotella. The type species is Prevotella melaninogenica.

During studies of human periodontal disease, a number of bacterial strains were encountered that, on the basis of results of standard biochemical tests, appeared to be Prevotella buccalis, Prevotella denticola, Prevotella melaninogenica, or Prevotella loescheii. However, use of the standard biochemical tests, cellular fatty acid analyses, and the polyacrylamide gel electrophoresis patterns of soluble proteins resulted in conflicting identifications of these strains. The results of tests for cellobiose fermentation, inulin fermentation, and pigment production were responsible for most of the discordant results. Cellular fatty acid analyses in which the Microbial Identification System was used did not differentiate these strains from validly described species, even though separate library entries were created for them. DNA reassociation determinations in which the S1 nuclease procedure was used showed that cellobiose fermentation and pigment production are variable among strains of P. melaninogenica and P. denticola and that fermentation of xylan is not a reliable characteristic for differentiating P. buccalis from Prevotella veroralis. In contrast to previous indications, most strains of P. veroralis do not ferment xylan. These species can be differentiated by DNA-DNA reassociation and by cellular fatty acid analysis, using the Microbial Identification System, but differentiation by currently described phenotypic characteristics is not reliable. Similarly, P. loescheii and the genetically distinct (but closely related) D1C-20 group cannot be distinguished reliably from each other or from P. veroralis, P. denticola, and P. melaninogenica on the basis of currently described phenotypic tests other than cellular fatty acid composition or, for some species, electrophoretic patterns of soluble whole-cell proteins.