Infants under age one in a pediatric practice were followed prospectively, and the determinants of acute gastrointestinal illness were evaluated in case-control pairs, matched by birth month. The risk of acute gastrointestinal illness in infants receiving formula was six times greater than in infants receiving breast milk and 2.5 times greater than in infants receiving cow milk. In the second six months of life, infants on formula had 0.38 more gastrointestinal illness episodes per child than infants on cow milk. Episodes without rotavirus or bacterial agents accounted for most of the increased risk of formula. The increased risk could not be explained by iron fortification of the formulas, prescription of non-milk based formulas to high-risk infants, case ascertainment bias, control selection bias, or numerous control factors. Non-antibody anti-infection properties found in cow milk are one possible explanation for these findings.