The association between breast feeding and acute lower respiratory infection (ALRI) was studied in a case-control study in southern Italy. Two groups of children were studied: the first group comprised 73 infants, aged 0-6 months, whose diagnosis was pneumonia or bronchiolitis; the second group included 88 infants less than 12 months of age with a diagnosis of pertussis-like illness. Control infants were two groups of infants admitted to the same ward. Compared with controls, infants in the first group were less likely to have been breast fed (odds ratio 0.42, 95% CI 0.19-0.90). The protection conferred by breast feeding was stronger among infants who were receiving human milk at the time of admission (odds ratio 0.22, 95% CI 0.09-0.55) and was absent among those infants who had stopped breast feeding for two or more weeks before admission. Among infants who were severely ill, breast feeding was less likely than among those with milder illnesses. There was evidence in the stratified analysis of effect modification by the presence of other children in the family. Among the infants with pertussis-like illness, the incidence and duration of breast feeding were not different compared with controls. The results suggest that breast feeding has a strong protective effect against ALRI in industrialized countries also. No protection seems to be conferred by human milk against pertussis-like illness.