Fetuses of mothers who have auto-immune thrombocytopenia are at risk in utero of having transitory thrombocytopenia. There is no maternal biological test that can predict the numbers of platelets in the fetus. This count of platelet numbers can be estimated from taking fetal blood from the scalp at the onset of labour, and if the count is less than 50 X 10(9) per litre a caesarean operation is indicated. The authors report a series of 23 pregnancies where the mother had auto-immune thrombocytopenia and where fetal scalp blood sampling was indicated. This series has been compared with a previous series of 14 pregnancies where scalp sampling was not carried out. There was no case of severe fetal thrombocytopenia after scalp sampling was carried out and where the best way of delivery could be predicted early enough in each case. The ease of taking the sample and the safety and reliability of it for estimating the degree of thrombocytopenia in the fetus and thus deciding the method of delivery makes us believe that this method should be used more frequently in cases of maternal thrombocytopenia that appear late in pregnancy.