We assessed the extent to which studies of the association between breast-feeding and infection met four important methodological standards that relate to both the scientific validity and the generalizability of the studies. Of the 20 studies (14 cohort, six case-control), only six met three or four of the methodological standards. Four of these six studies found that breast-feeding was not protective against infections and two found that breast-feeding was protective against infections. In the three studies in which statistical adjustments were made for three additional potentially important confounding variables--size of the family, smoking of cigarettes by the mother, and the mother's level of education-the apparent protective effect of breast-feeding against respiratory tract infections disappeared after the adjustments were made. We found that most of the studies have major methodological flaws that may have compromised their conclusions. The studies that met important methodological standards and controlled for confounding variables suggest that breast-feeding has at most a minimal protective effect in industrialized countries.