To assess the association between infant feeding and childhood cancer, a qualitative review of 9 published case-control studies was undertaken. The results of this synthesis suggest that children who are never breast-fed or are breast-fed short-term have a higher risk than those breast-fed for > or = 6 months of developing Hodgkin's disease (HD), but not non-Hodgkin's lymphoma or acute lymphoblastic leukemia. HD has features of a complex cellular immune disorder and of chronic infection. Human milk contains an extensive array of anti-microbial activity and appears to stimulate early development of the infant immune system. Artificially fed infants negotiate exposure to infectious agents without the benefits of this immunologic armament and do not do as well as breast-fed infants in resisting infection. Thus, human milk may make the breast-fed infant better able to negotiate future carcinogenic insults by modulating the interaction between infectious agents and the developing infant immune system or by directly affecting the long-term development of the infant immune system. Further research should attempt to confirm the association between infant feeding and HD in large, population-based, case-control studies. Improved measurement of infant feeding must be addressed if future studies are to advance our understanding of this association. In addition, studies of specific measures of immunity, particularly of cellular immune responses, should be conducted in populations of breast-fed and non-breast-fed young children.