Fifty-nine fetuses of severely sensitized rhesus-negative women were selected for intrauterine fetal transfusion. This selection was based primarily on the level of amniotic fluid pigment and previous obstetric history; serum antibody titers were found to be of little or no value. Thirty of the 59 fetuses were stillborn; 13 apparently died as a direct result of intrauterine fetal transfusion, although lethal trauma was found in only 3. There were 10 neonatal deaths, most of which were due to hydrops fetalis. Nineteen infants survived, a salvage rate of 32.2 per cent. The influences on fetal survival of amniotic fluid pigment level, the presence of intrauterine hydrops fetalis, gestational age at the time of the initial intrauterine fetal transfusion, and the time and type of delivery are discussed. The salvage of one third of infants whom we believe would have died otherwise seems to both justify and demand intrauterine fetal transfusion.