The classical twin model was utilized in this study in an attempt to determine the importance of host genetics to the composition of the subgingival flora. Simultaneously, the effect of puberty on the flora composition was assessed. The compositions of the floras were significantly different at ages 11 and 14 in the same people, indicating that transition to an adult flora composition may be initiated during puberty. However, the numbers of subjects who had prepubertal and postpubertal testosterone levels in this study were too small to demonstrate significant differences based solely on testosterone level (P = 0.053 and 0.11 for tests of unrelated members, i.e., all twins "a," the first twin of each pair, and all twins "b," the second twin of each pair). Sixteen unrelated 11-year-old subjects had prepubertal levels of less than 30 ng of testosterone per dl of serum, and only six of these unrelated subjects had levels above 300 ngldl by age 14. Of their twin siblings, who formed the second group of unrelated individuals, 15 had prepubertal levels and only 5 reached postpubertal levels. Unpaired t tests indicated that VeilloneUla atypica, PrevoteUla denticola, and PrevoteUla melaninogenica were among the species that contributed most to changes in flora composition during puberty. The compositions of subgingival floras of 11-year-old monozygous and dizygous male twins were significantly more similar than those of unrelated subjects in the study (P = 0.004 and 0.009, respectively). At 12.5 years of age, the floras of monozygous twins remained more similar than those of unrelated subjects (P = 0.001), but the dizygous-twin floras were not significantly more similar than those of unrelated people. This difference corresponded with moderate and varied testosterone levels within dizygous-twin pairs at age 12.5. By age 14 both monozygous and dizygous twins again had floras with compositions more similar than those of unrelated people (P = 0.008 and 0.002, respectively). Estimates of the genetic contributions to the increased similarity of the floras of twins as compared with floras of unrelated people indicated that the concentrations of several species in the flora may be influenced by host genetic factors. The prevalence of certain other species appeared to be controlled