HLA-DR/DQ-DP linkage disequilibrium was investigated in healthy, unrelated British (n = 150) and French Canadian (n = 67) Caucasoid subjects. HLA-DR and DQ typing was performed by Taq I DNA-RFLP analysis, while DPB1 typing was performed by PCR-SSOP. chi 2 and Fisher's exact tests were performed for all 2-locus biallelic comparisons and coefficients of linkage disequilibrium determined. In the British population, only one example of linkage disequilibrium, significant at P = 0.05 (after correction for the number of comparisons made) was seen (DPB1*0101-DRB1*0301[17(1)]). Additional associations, significant at P = 0.05 before correction for the number of comparisons were also seen, including DPB1*0401-DRB1*15, DPB1*1101-DRB1*0701(7(1)), DPB1*1701-DRB1*0701/ 2(7(2)), DPB1*0101-DQA1*0501, DPB1*0401-DQA1*0102, DPB1*0501-DQA1*0102, DPB1*0101-DQB1*0201, DPB1*0401-DQB1*0602/0603 and DPB1*1101-DQB1*0201. With one exception (DPB1*1101-DQB1*0201), none of these associations was seen in the French Canadian group. These results indicate that although more frequent than thought hitherto, HLA class II linkage disequilibrium involving DPB1 alleles is generally weak, and can differ even between different caucasoid populations. This may have implications for HLA and disease studies.