The consistency of colonization of tooth surfaces by Streptococcus mutans was studied by sampling four approximal molar surfaces of 9- to 13-year-old children at weekly intervals. It was found that surfaces which harbored detectable levels of the organism tended to be consistently positive, whereas surfaces which were free of detectable levels of the organism also tended to remain that way. Of the total surfaces studied, 81.7% were either consistently positive or consistently negative throughout five consecutive weekly samplings. These data indicate that S. mutans does not uniformly colonize the surfaces of teeth. The consistent localized patterns of colonization observed suggest that S. mutans is not readily transmitted from one tooth surface to another within the mouth. Scanning electron microscope observations revealed that S. mutans colonized the teeth of monoinfected gnotobiotic rats as discrete localized colonies, whereas strains of Actinomyces naeslundii colonized in a more generalized manner. Efforts to sterilize tooth surfaces by application of iodine solution indicated that the populations of S. mutans on some surfaces could be affected for at least 11 to 13 weeks after treatment. The number of tooth surfaces colonized by detectable proportions of S. mutans was found to be far lower in a group of 18 children from Charlotte, N.C., with a low past caries experience than in 20 children from Danvers, Mass., with a high caries experience. The number of tooth surfaces infected with S. mutans therefore strongly paralleled the caries experience of these populations.