OBJECTIVE: The aim of this study was to determine the accuracy of noninvasive fetal RhD genotyping by fetal cell isolation from maternal blood. STUDY DESIGN: Candidate fetal cells from 18 pregnant women (one twin gestation) were flow-sorted. Polymerase chain reaction amplification of a 261 bp fragment of the RhD gene was performed on sorted fetal cells. The presence of amplified product was considered predictive of the RhD-positive genotype in the fetus. RESULTS: Sixteen of the 19 fetal RhD genotypes were correctly predicted in fetal cells isolated from maternal blood (10 were Rh positive, 6 were Rh negative). In 3 cases no amplification products were detected in RhD-positive fetuses. The association between presence of the fragment and RhD-positive genotype was significant (p = 0.003, Fisher's exact test). CONCLUSIONS: Noninvasive prenatal diagnosis of the fetal RhD genotype is feasible. Absence of amplification products in the reaction requires confirmation that fetal material is present. Improvements in fetal cell purity and yield should increase diagnostic accuracy, although the current protocol has a positive predictive value of 100% and a negative predictive value of 67%. (AM J OBSTET GYNECOL 1996;174:818-22.)