Aluminium was measured in samples of plasma and samples of feed obtained from 74 infants with normal renal function established on various feeds (breast, whey-based, fortified whey-based, preterm, soy, and casein hydrolysate). All infants were bolus fed, and blood samples were collected midway between feeds. Aluminium was measured using electrothermal atomization and atomic absorption spectrometry. Mean aluminium concentrations in milks were as follows: breast, 9.2 micrograms/L [95% confidence interval (CI), 5.6-12.7]; whey-based, 165 micrograms/L (95% CI, 151-180); fortified, 161 micrograms/L (95% CI, 143-180); preterm, 300 micrograms/L (95% CI, 272-328); soy, 534 micrograms/L (95% CI, 470-598); casein hydrolysate, 773 micrograms/L (95% CI, 632-914). Mean plasma aluminium concentrations in infants receiving different milks were as follows: breast, 8.6 micrograms/L (95% CI, 5.6-10.6); whey-based, 9.2 micrograms/L (95% CI, 7.4-11.0); fortified, 10.3 micrograms/L (95% CI, 8.3-12.3); preterm, 9.7 micrograms/L (95% CI, 5.3-17.1); soy, 12.5 micrograms/L (95% CI, 5.0-20.0); casein hydrolysate, 15.2 micrograms/L (95% CI, 10.7-19.8). Mean plasma aluminium concentration was significantly different in infants fed casein hydrolysate formulae than in those fed breast milk (difference, 6.7 micrograms/L; 95% CI, 2.8-10.5; p = 0.028). We conclude that infants may be at risk from aluminium toxicity when consuming formula containing > 300 micrograms/L--in particular, casein hydrolysate formulae. We speculate that the aluminium compounds found in breast milk are more bioavailable than those found in other milks and that some constituents of infant formula affect aluminium absorption from the gut lumen.