The records of 32 neonates in an intensive care unit were examined retrospectively to determine if fetal hemoglobin concentrations could be predicted on the basis of gestational or postnatal age, or on the volume of red blood cell transfusions. In nontransfused neonates, the correlation between measured concentrations of fetal hemoglobin and post-natal age was r = 0.53 with a 17.2 standard error of prediction. In these same neonates, the correlation between measured fetal hemoglobin divided by birth weight and gestational age was r = 0.70, with a 9.6 standard error of prediction. A three-variable regression equation (the latter two variables plus calculated fetal hemoglobin) was found to have a high correlation with data for measured fetal hemoglobin (r = 0.97) and a relatively low 8.4 standard error of prediction. In transfused neonates, however, measured hemoglobin concentrations divided by birth weight correlated poorly with gestational age (r = 0.30 and a 12.4 standard error of prediction). In addition, the transfused neonates had low correlations when fetal hemoglobin concentrations alone were compared with the total volume of red blood cell transfusions (r = 0.35) and with postnatal age (r = 0.18) and the standard errors of prediction were all approximately 17. The correlations found between concentrations of fetal hemoglobin and age in transfused neonates were poorer than those reported in earlier nontransfused infant studies. Previous studies have also shown that neonatal blood containing fetal hemoglobin interferes with the spectrophotometric measurements of carboxyhemoglobin and oxyhemoglobin. Because of the imprecision in the predictions of fetal hemoglobin using age, weight, or the volume of transfusion, we conclude that fetal hemoglobin should be measured if accurate spectrophotometric determinations of carboxyhemoglobin and oxyhemoglobin are desired.