During the last ten years several workers ^1,3,5,6 have made use of large series of mother-child blood group combinations to infer either the presence or absence of selection in the ABO blood group system. Such studies depend for their final validity on the accuracy of the determination of the blood groups of the mothers and of their newborn children. Routine typing of adult blood involves checks to avoid clerical errors, and the use of reverse grouping to avoid misclassification due to false positive or negative cell agglutinations ^9,2. Clerical checks can remove this source of error in the routine typing of the blood of newborn infants but the absence of natural anti-A or anti-B antibodies in the serum of newborn infants, except those derived by placental transfer from the mother, precludes the use of reverse grouping. This is particularly unfortunate since it is now well established that the average strength of the A or B antigens of the erythrocytes of newly bom children is less than that of adults ^10,4,8. The weaker antigenic strength of such erythrocytes might result in a larger number of false negative reactions being scored, even by experienced serologists. In order to check the magnitude of such errors in our own routine cord-blood typing we have made use of the fact that the secretor status of newborn infants can be determined from saliva samples collected from the babies during the first few days after birth ^11,13. Not all errors can be detected by this method, but false negative results should be revealed in any secretor baby whose red cell A or B antigen has been missed.