To test the hypothesis that enhanced intestinal absorption of bilirubin may contribute to prolonged nonconjugated hyperbilirubinemia in human milk-fed infants, we studied a cross-section of 36 healthy infants and mothers. Milk from mothers and serum from infants were collected at 16.3 +/- 2.4 days. Milk was studied for its effect on the absorption of bilirubin labeled with carbon 14 in rats and compared with buffer and iron-fortified infant formula (Similac With Iron). The percentage of a 1 mg bilirubin dose absorbed by the rat was 25.29 +/- 4.0% when it was administered into the duodenum with buffer, 4.67 +/- 2.4% with Similac formula, and 7.7 +/- 2.9% with human milk. Linear regression analysis, using the infant's serum nonconjugated bilirubin level as the dependent variable and the percentage of (14C)bilirubin absorbed by the rat with the corresponding mother's milk as the independent variable, revealed a significant correlation (r = 0.40; p = 0.016). Inspection of the data suggested that absorptive permissiveness correlated closely with infant serum bilirubin values greater than 24 mumol/L (1.4 mg/dl) (r = 0.55; p = 0.007), whereas in those with bilirubin values less than or equal to 24 mumol/L, there was no apparent correlation. Milk was also analyzed for beta-glucuronidase, nonesterified fatty acids, and the ability to inhibit glucuronosyltransferase activity of rat liver microsomes in vitro, none of which correlated with the infant's serum bilirubin. These data support the theory that enhanced intestinal absorption of bilirubin contributes to the jaundice associated with breast-feeding.