At birth and at 4 months of age, selenium (Se) values of 129 term infants on three different diets were determined: 50 infants were breast fed (HM), 44 received formula based on cow's milk (F) and 35 were fed "hypoallergenic formula" (PHF) (partially hydrolysed whey protein). The Se status of a group of twins (n = 12) fed "hypoallergenic formula" was compared with the respective group of singletons. All infants had low plasma Se values during early infancy. The plasma Se of breast-fed infants remained stable (plasma Se 43 +/- 8 ng/ml at birth and at 4 months), whereas plasma glutathione peroxidase (GSH-Px) decreased (birth: 107 +/- 29 U/l; 4 months: 62 +/- 11 U/l). The formula-fed infants showed a reduction in plasma Se levels from birth to 4 months (38 +/- 10 ng/ml and 29 +/- 9 ng/ml, respectively). The decrease was even more pronounced in infants fed the "hypoallergenic formula". This group presented the lowest Se values (plasma Se 39 +/- 9 ng/ml at birth; 20 +/- 6 ng/ml at 4 months). Renal excretion of Se was found to be lower in the formula-fed infants (F and PHF) compared with the HM group. There was a significant correlation between plasma and urinary Se (r = 0.62, p = 0.0001). Urinary Se (microgram Se/g creatinine) appeared to be a good indicator of Se intake. Measurements of urine Se might be used as a screening method for the estimation of the Se supply. Weight and length increases in all infants were within the normal range. There were no differences between the different feeding groups.