The HLA-B41 specificity was first identified over 25 years ago and, although both serological and biochemical studies have suggested its subdivision, it is only recently that two HLA-B*41 alleles (B*4101 and B*4102) have been identified and sequenced. We designed three oligonucleotide primers, combined in two mixtures to define these alleles by PCR using sequence-specific primers (PCR-SSP) in a random normal population of 9,464 HLA-A, B, DR, DQ typed Northern European Caucasoid donors from the Welsh Bone Marrow Donor Registry. The HLA-B41 phenotype frequency was 0.835%, and of the 79 HLA-B41 subjects 22 (27.85%) were B*4101 and 57 (72.15%) were B*4102. The phenotype frequencies of B*4101 and B*4102 were 0.232 and 0.602%, respectively, and the gene frequencies were 0.00116 and 0.00301, respectively. Formal two-locus linkage disequilibrium (LD) analysis demonstrated significant associations between B*4101 and A30 and DR11, and between B*4102 and A66 and DR13. LD analysis of three loci showed significant associations between B*4101, DR7, DQ2 and B*4101, DR11, DQ7 (DQB1*0301/0304) and between HLA-A3, B*4102, DR13; A66, B*4102, DR7; A66, B*4102, DR13; B*4102, DR13, DQ6 and B*4102, DR13, DQ7. Examination of the HLA phenotypes (including HLA-C) of the B*41 subjects, together with the LD analysis findings, suggested four different HLA haplotypes bearing B*4101 and five B*4102 haplotypes. The most frequent B*4101 haplotype was: HLA- A30 or other A allele, Cw*1701, B*4101, DRB1*1102, DQB1*0301 and the most freqent B*4102 haplotype was: A*6601 or A3 or other A allele, Cw*1701, B*4102, DRB1*1303, DQB1*0301. In addition, the well-known association of A66 with B41 was between A*6601 and B*4102, and although both B*41 alleles were commonly in association with Cw*1701, B*4101 was found with Cw*07. One-dimensional isoelectric focusing (1D-IEF) analysis of HLA-B proteins from 2 B*4101 and 2 B*4102 subjects clearly showed that the B41.1 and B41.2 1D-IEF variants, identified in the 10th International Histocompatibility Workshop, corresponded to B*4101 and B*4102 products, respectively. Serological titration studies, with 59 lymphocytotoxic pregnancy antisera, containing an HLA-B41 component and stimulated by up to five different HLA-B specificities, were unable to differentiate the two groups of B*41 subjects. This suggests that partition of the HLA-B41 specificity will not normally be achieved by classical serological methods. It is suggested that the previous alleged serological subdivision of HLA-B41 was founded on the unwitting use of antisera detecting the HLA-Cw*17 products.