Screening groups of anonymous infants for HIV antibody, as an index of maternal infection rates, has been a widely used seroepidemiological method since being introduced in 1986 in Massachusetts (USA). One shortcoming has been the applicability only to parturient women, thus necessitating corrections for fertility rates in extrapolation to all women. A second disadvantage has been controversy and confusion about the distinction between anonymous seroprevalence studies and linked testing. However, there have been major advantages such as the "leading indicator" nature of the data obtained. Experience with screening nearly a half million Massachusetts newborns through December 1992 has shown seroprevalence rates stabilizing at 2.4 per 1000, and with consistent 10-fold differences between groups of birth hospitals serving different socio-economically defined populations. In addition to predicting the future of the AIDS epidemic in children, the information provides a reference point for comparing the completeness of targeted identifications of HIV infection in mothers and infants.