Pulsed Doppler studies of the fetal and maternal circulations were carried out before and after 64 intrauterine transfusions performed on 24 fetuses. A model was derived for the prediction of hematocrit before the first transfusion: Hematocrit = 7.778 - (0.088 x peak velocity in descending aorta) + (0.968 x gestational age [weeks]) - (10.911 if hydrops present) (r = 0.876, p less than 0.0001). An alternative formula, excluding hydrops, was slightly less predictive: Hematocrit = 45.312 - (56.261 x umbilical cord Pourcelot index) - (0.128 x peak velocity in descending aorta) + (1.042 x gestational age) r = 0.822, p less than 0.001). Neither model was accurate in the prediction of hematocrit before second or subsequent transfusions. A third model was derived from second-transfusion data: Hematocrit = 40.524 - (0.045 x peak velocity in descending aorta) - (10.693 x pulsatility index of maternal uterine artery) (r = 0.81, p less than 0.003). However, this model was unable to predict hematocrit before third or later transfusions. No changes in Doppler parameters before and after transfusion were found. We conclude that pulsed Doppler ultrasound may be helpful in the evaluation of isoimmunized pregnancies, in differentiating anemic from normal fetuses. It does not appear to be useful in determining the timing of later transfusions. The lack of change before and after transfusions suggests that these vessels will not provide significant information concerning the effect of rapid volume and hematocrit changes in the fetus.