Screening for atopic disease using neonatal serum IgE has been advocated on the basis of the predictive value of elevated levels. However, this is only one measure of validity. The test was validated fully in 92 infants with a bi-parental history of atopic using 0.7 IU/ ml as the cut-off. All infants were assessed prospectively for evidence of atopic disease (eczema, recurrent wheezing or food reactions) and skin-prick test positivily in the first year. Total serum IgE was measured by ultrasensitive ELISA on 61 cord blood samples and 92 samples taken at 7 days. All cord samples were re-analysed by PRIST and the first 33 by ultrasensitive RIA giving, respectively, 82% and 94% concordance (regarding undetectable, detectable and elevated levels) with ELISA. Maternal contamination was indicated in 7% of cord samples by high serum IgA. Ninety-five per cent of cord/7-day IgE pairs showed no change or minor rises at 7 days. Forty-nine per cent of the infants had evidence of atopic disease. Only 5% had elevated 7-day IgE. The positive and negative predictive values of the 7-day test were 60% and 52%, respectively, and specificity 96% but the sensitivity was only 7%. High levels did not distinguish the infants with the most unequivocal evidence of disease, i.e. eczema with a positive skin test. In conclusion IgE at 7 days is comparable to and more reliable than cord IgE. However, neonatal IgE screening is too insensitive to have clinical application.