The aim of this study was to determine the fetal loss rate after fetal blood sampling (FBS) in relation to the indication. In total, 1981 FBS procedures (1878 pregnancies) were included, of which 117 were performed for the detection of congenital infection (group 1), 1437 for the detection of haemoglobinopathy (group 2), 233 for prenatal diagnosis with normal ultrasound findings (group 3), 121 for rapid karyotyping in cases with abnormal sonographic findings (group 4) and 73 for severe growth retardation (group 5). All the procedures were performed with a free-hand technique under continuous ultrasound guidance. Pregnancy losses occurring within two weeks of FBS were considered procedure-related losses. 343 pregnancies were terminated. Of the remaining 1535 continuing pregnancies, 73 (4•8 per cent) were lost, of which 39 (2•5 per cent) were lost within two weeks of the procedure. The procedure-related losses were 3 in 103 (2•9 per cent), 17 in 1090 (1•6 per cent), 2 in 191 (1 per cent), 11 in 84 (13•1 per cent) and 6 in 67 (8•9 per cent) in groups 1, 2, 3, 4 and 5, respectively. The differences in procedural loss between the five groups were highly significant, suggesting that the method entails a much higher risk when the fetus is structurally abnormal, or severely growth retarded. Patients should therefore be counselled before the procedure accordingly.