Since AIDS was first described in children, much has been learnt about the epidemiology, clinical presentations and natural history of perinatally acquired HIV disease. As yet, several questions remain unanswered about the pathogenesis of vertical transmission, the relative risk associated with each mode of transmission, the rate of transmission from mother to child, and the factors that might contribute to the efficiency of transmission. Data presented thus far suggest that intrauterine exposure poses the greatest risk, but more recent reports have reopened speculation that infection could occur at or around the time of delivery. The prospects for intervention, to stop transmission from mother to child, rest on the ability to identify HIV-infected women; the knowledge of how and when the virus infects the fetus; and how to identify, as early as possible, those truly infected infants. Resources should therefore be set aside for routine screening for HIV in antenatal women. There is also a need to quantify the contribution made by intrauterine versus intrapartum events. Lastly, promising new techniques designed to detect neonatal HIV infection must be properly evaluated against standard methods, and correlated with clinical outcome. Only then can the efficacy of antiretroviral therapy be tried, to prevent vertical transmission.