The association of breast feeding with rates of atopic illness during the first five years of life was assessed in a national study of 13 135 children studied during the first week and at age 5 years. Eczema was reported more often in children who had been breast fed; this relationship persisted even after allowance was made for social and family factors influencing the likelihood both of breast feeding and of eczema; the other factors most significantly associated with rates of eczema were parental history of eczema or asthma and advantaged family socioeconomic status. A similar, but less pronounced, positive association of breast feeding with reported hayfever became non-significant after adjustment for intervening factors. Rates of reported asthma were not influenced by breast feeding. "Any wheezing" including asthma was reported more often in children who had not been breast fed, but this association disappeared after adjustment for parental asthma and maternal smoking. Breast feeding does not appear to protect against these atopic diseases. The positive association with reported eczema might relate to accuracy of diagnosis or to associated influences not considered in the analysis; alternatively, it might be due to (recent) environmental contaminants crossing in breast milk, causing eczema in the child.