Human fetal cells in maternal blood were analyzed using special high-speed (>100,OOO cells/sec), three-color fluorescence flow cytometry.'-' High-resolution sorting4 of fetal cells was also performed, and sorted single cells were characterized by polymerase chain reaction (PCR) (FIG. 1). Fetal cell subpopulations were identified by three different methods. In the fust method, erythrocytes in maternal whole blood were gently lysed, then stained with F'ITC-conjugated anti-CD7 1 monoclonal antibody and PE-conjugated anti-glycophorin-A (GPA) monoclonal antibody. Subsequently, the cells were fixed overnight in 1% paraformaldehyde, permeabilized for 15 min at room temperature with 0.1% Triton-X1 00, and then stained with the nuclear stain, 7-aminoactinomycin D (7AAD). The putative CD71+GPA7AALlt nucleated fetal erythroid cells were then examined by three-color fluorescence flow cytometry. Then putative rare cells were sorted directly into PCR tubes at lower speeds for confirmation of fetal cell identity by subsequent PCR analysis and detection of HLA-DQa alleles using multiplex PCR and dot-spot hybridizations with allele-specific oligonucleotide probes. Because sorts of 10 cells yielded, at best, fetal cells contaminated with maternal cells, high-resolution single-cell, multi-tube PCR sorting was performed to try to obtain pure single fetal cells for molecular analyses. There has been initial success with this new approach. A second method used was a flow cytometric analysis of whole unlysed maternal blood using a triple antibody-labeling scheme with positive (CD71 and CD34) and negative (CD45 negative and dim) selection markers (cf. FIG. 2). A subpopulation of putative fetal cells at a frequency of 1.8 x (with a 95% probability of this subpopulation being between 1.8 x and 2.0 x 10.' by finite sampling statistics) was detected as shown in FIGURE 2B.