BACKGROUND: Breast-feeding is well known to have a protective effect against infection in infants. Although the long-term effects of breast-feeding on childhood cancer have not been studied extensively, a protective effect against childhood Hodgkin's disease and lymphoma has been suggested previously from small investigations. In this study, we tested the hypothesis that breast-feeding decreases the risk of childhood acute leukemia. METHODS: A total of 1744 children with acute lymphoblastic leukemia (ALL) and 1879 matched control subjects, aged 1-14 years, and 456 children with acute myeloid leukemia (AML) and 539 matched control subjects, aged 1-17 years, were included in the analysis. Information regarding breast-feeding was obtained through telephone interviews with mothers. All leukemias combined, histologic type of leukemia (ALL versus AML), immunophenotype of ALL (early pre-B cell, pre-B cell, or T cell), and morphology of AML were assessed separately in the data analysis. RESULTS: Ever having breast-fed was found to be associated with a 21% reduction in risk of childhood acute leukemias (odds ratio [OR] for all types combined = 0.79; 95% confidence interval [CI] = 0.70-0.91). A reduction in risk was seen separately for AML (OR = 0.77; 95% CI = 0.57-1.03) and ALL (OR = 0.80; 95% CI = 0.69-0.93). The inverse associations were stronger with longer duration of breast-feeding for total ALL and AML; for M0, M1, and M2 morphologic subtypes of AML; and for early pre-B-cell ALL. CONCLUSION: In this study, breast-feeding was associated with a reduced risk of childhood acute leukemia. If confirmed in additional epidemiologic studies, our findings suggest that future epidemiologic and experimental efforts should be directed at investigating the anti-infective and/or immune-stimulatory or immune-modulating effects of breast-feeding on leukemogenesis in children.