BackgroundTo better understand potential transmission risks from contact with the body fluids of children, we monitored the presence and amount of CMV shedding over time in healthy CMV-seropositive children.MethodsThrough screening we identified 36 children from the Atlanta, Georgia area who were CMV-seropositive, including 23 who were shedding CMV at the time of screening. Each child received 12 weekly in-home visits at which field workers collected saliva and urine. During the final two weeks, parents also collected saliva and urine daily.ResultsPrevalence of shedding was highly correlated with initial shedding status: children shedding at the screening visit had CMV DNA in 84% of follow-up saliva specimens (455/543) and 28% of follow-up urine specimens (151/539); those not shedding at the screening visit had CMV DNA in 16% of follow-up saliva specimens (47/303) and 5% of follow-up urine specimens (16/305). Among positive specimens we found median viral loads of 82,900 copies/mL in saliva and 34,730 copies/mL in urine (P = 0.01), while the viral load for the 75th percentile was nearly 1.5 million copies/mL for saliva compared to 86,800 copies/mL for urine. Younger age was significantly associated with higher viral loads, especially for saliva (P < 0.001). Shedding prevalence and viral loads were relatively stable over time. All children who were shedding at the screening visit were still shedding at least some days during weeks 11 and 12, and median and mean viral loads did not change substantially over time.ConclusionsHealthy CMV-seropositive children can shed CMV for months at high, relatively stable levels. These data suggest that behavioral prevention messages need to address transmission via both saliva and urine, but also need to be informed by the potentially higher risks posed by saliva and by exposures to younger children.Electronic supplementary materialThe online version of this article (doi:10.1186/s12879-014-0569-1) contains supplementary material, which is available to authorized users.