The relation between maternal anti-D concentrations, measured against the British working standard, and outcome of rhesus-sensitised pregnancies was studied. There is a clear relation between increasing anti-D concentrations and the chance of a severely affected baby. Of those pregnancies (78) where serial anti-D concentrations remained below 4 IU/ml, no baby had a cord haemoglobin below 10 g/dl and three had exchange transfusions. In contrast, of those mothers (106) with anti-D concentrations above 4 IU/ml, 23 had babies with a cord haemoglobin below 10 g/dl and 79 babies had exchange transfusions. It is suggested that those pregnancies where anti-D concentrations remain below 4 IU/ml represent a relatively safe group in which amniocentesis may be avoided.