Umbilical cord blood is considered an alternate source of hematopoietic stem cells in bone marrow transplantation. However, its use might be hampered by contamination of neonatal blood with maternal cells, which could contribute unacceptably to graft-vs.-host disease (GVHD) after transplant. In a previous study (Socié et al., Blood 83:340, 1994), we used polymerase chain reaction (PCR) amplification of minisatellite sequences (sensitivity 1-0.1%) to address the question of this contamination. In a single case among 47 analyzed, we were able to detect a maternal-specific allele in the cord blood sample. We have now studied the same cord samples using a highly sensitive, allele-specific PCR amplification method. A maternal allele could be discriminated from neonate alleles in 10 cases and maternal cells were detected in all 10 cord blood samples. These cells amounted to 10(-4) to 10(-5) of cord blood nucleated cells. In three cases, cord blood separated cell subpopulations could be analyzed and were found to contain maternal cells at about the same level. The presence of maternal cells at such a low level in cord blood samples probably would have no effect on GVHD in a clinical setting of transplantation but raises interesting questions in terms of materno-fetal immune tolerance and transmission of viruses (in particular human immunodeficiency virus) from infected mother to child.