We developed a method for polyacrylamide gel electrophoresis of whole cells of Actinomyces spp. The major advantage of this methad is its ease of operation since it obviates the need for the preparation of bacterial extracts. Whole-cell samples were prepared by incubating washed, packed cells in 6 M urea at 37°C for 24 h. The results of disc gel electrophoresis of these whole-cell samples were compared with the results of electrophoresis of soluble protein extracts of the bacteria run under the same conditions. A large number of bands were obtained with the whole-cell preparations, and these bands were resolved better than bands obtained with the protein extracts. A total of 22 strains of Actinomyces spp. were examined by polyacrylamide gel electrophoresis of urea-treated whole cells. A cluster analysis of the polyacrylamide gel electrophoresis band patterns showed that the strains clustered first according to serotype and then according to species. Overall, two main divisions were identified, one containing Actinomyces bovis, Actinomyces odontolyticus, and Actinontyces israelii and one containing Actinomyces viscosus and Actinomyces naeslurtdii. This method should be a valuable tool in studying the taxonomy of Actinomyces spp.In Bergey 's Manual of Determinative Bacteriology, 8th ed. (181, the genus Actinomyces is divided into five species. Since the publication of Bergey's Manual, the species Corynebacterium pyogenes has been transferred to Actinomyces (9, the specific epithet Actinomyces meyeri has been revived (3), and three new animal species, Actinomyces denticolens (7), Actinornyces howellii (8), and Actinomyces hordeovulneris (2), have been described. A number of problems in the classification of Actinomyces remain. Many normal flora isolates from oral cavities cannot be placed in a species and several studies raise questions concerning some of the species divisions given in Bergey's Manual.Two numerical taxonomy studies (10, 16) suggest that Actinomyces israelii may deserve recognition as a separate genus. These studies and others (6, 9) raise questions concerning the species Actinomyces naeslundii and Actinomyces viscosus. A. naeslundii and A . viscosus are closely related phenotypically, demonstrating about the same degree ofrelatedness in numerical studies as the two serotypes of A. israelii do (10). Rodent strains of A. viscosus are different from human isolates serologically and in some other ways. Simply combining A. naeslundii and A. viscosus into a single species does not appear to solve the problem since subclusters within each species occur in numerical studies (9,10,16). Deoxyribonucleic acid (DNA)-DNA homology studies support division of A. naeslundii and A . viscosus into two species and separation of rodent and human isolates of A. viscosus (6).Polyacrylamide gel electrophoresis (PAGE) could prove to be valuable in answering these taxonomic questions. PAGE of soluble proteins can be used to differentiate closely related bacteria (1,11,13,17) and has been used to show differences among stra...