Breast-feeding policy tends to be an emotive issue. International agencies recommend exclusive breast-feeding for 4-6 months followed by continued partial breast-feeding into the second year of life in order to promote infant and child health and minimize the damage caused by the malnutrition-infection cycle. To what extent are these recommendations supported by the experimental evidence? Are they a simplification for emotional reasons or public health purposes? Breast-feeding is believed to benefit infants because breast milk contains the ideal mix of nutrients for infants, because it contains factors which promote development of the infant's gut and immune system and which prevent pathogen invasion, and because exclusive breast-feeding prevents intake of pathogens in food or water. However, some apparently contradictory evidence exists. First, in environments which are not highly contaminated breast-fed infants tend to growth falter relative to those fed formula. Second, in such environments partial breast-feeding is not associated with significantly increased gut damage relative to exclusive breast-feeding, suggesting that active promotion of gut development by breast-feeding is more important than simple avoidance of pathogens from other foods. Third, many immune factors in breast milk are probably present primarily to protect the mother, not the infant. Finally, breast milk itself may contain bacteria or viruses. This problem has come to the fore with the human immunodeficiency virus epidemic, since it is clear breast-feeding is an important mode of mother-to-child transmission. The present review will examine these challenges to the basis of the international infant feeding recommendations and will suggest that the science does actually support the policy.