The huge majority of the annual 6.3 million perinatal deaths and half a million maternal deaths take place in developing countries and are avoidable. However, most of the interventions aiming at reducing perinatal and maternal deaths need a health care system offering appropriate antenatal care and quality delivery care, including basic and comprehensive emergency obstetric care facilities. To promote the uptake of quality care, there are two possible approaches: influencing the demand and/or the supply of care. Five lessons emerged from experiences. First, it is difficult to obtain robust evidence of the effects of a particular intervention in a context, where they are always associated with other interventions. Second, the interventions tend to have relatively modest short-term impacts, when they address only part of the health system. Third, the long-term effects of an intervention on the whole health system are uncertain. Fourth, because newborn health is intimately linked with maternal health, it is of paramount importance to organise the continuum of care between mother and newborn. Finally, the transfer of experiences is delicate, and an intervention package that has proved to have a positive effect in one setting may have very different effects in other settings.