To evaluate the clinical significance of eosinophilia in newborn infants, 261 admissions to the neonatal unit over a 12-month period were studied retrospectively; 33 babies with eosinophilia (> 1.0 x 10(9)/l) were studied. Clinical and laboratory data for the first month of life were compared, where available, between gestational age-matched pairs with and without eosinophilia. Of the 33 babies with eosinophilia, 23 were > 26 weeks' gestation and all had age-matched controls; 10 were < or = 26 weeks' gestation but had no appropriate gestational age-matched controls. Babies > 26 weeks' gestation with eosinophilia had a significantly higher number of septic episodes than controls: 20 of 23 versus 4 of 23. All 10 babies < or = 26 weeks' gestation with eosinophilia developed sepsis. Infections with gram-negative organisms and necrotizing enterocolitis occurred only in babies who developed eosinophilia. In 5 babies no cause for the eosinophilia was found. In conclusion, eosinophilia in the newborn is usually explainable and is most often associated with infection.