A total of 42,160 individuals were typed for HLA-A and HLA-B by both serology and PCR-based typing. The HLA assignments included all of the known serological equivalents. The majority of the individuals (99.9%) were from U.S. minority population groups. The serologic typing was performed between 1993 and 1997 at the time of recruitment for the National Bone Marrow Program (NMDP) registry. The polymerase chain reaction (PCR)-based typing was carried out in two phases. In phase I, DNA typing was performed by PCR using sequence-specific oligonucleotide probes (PCR-SSOP) or PCR using sequence-specific primers (PCR-SSP) without knowledge of the serologic assignments. Discrepancies were identified between the serologic and DNA assignments in 24% of the volunteers (8% of volunteers differed for only HLA-A assignments, 13% for HLA-B, and 3% for both HLA-A and -B) and a potential explanation was assigned each discrepant serology/DNA pair. In phase II, a random sampling scheme was used to select a statistically significant number of individuals for repeat DNA typing from each of these categories. The categories included antigens missed by serology, nonexpressed (null) alleles, PCR amplification failures, misassignment of antigens and nomenclature issues. Only a single individual was found to carry a null allele. DNA-based testing correctly typed nearly 99% of the donors at HLA-A, more than 98% at HLA-B, and more than 97% at both HLA-A and -B validating this methodology for registry typing.