The association of particular HLA-DR alleles and the shared epitope with rheumatoid arthritis (RA) is now well established. The strength of these links varies between races. Furthermore, the proposition that the presence of the shared epitope is indicative of severe disease has been more difficult to sustain in non-Europeans. This study examines the frequency of HLA-DR and HLA-DRB1 amongst Pakistanis for the first time. Using the polymerase chain reaction (PCR) and sequence-specific oligonucleotide probes (PCR-SSOP) and primers (PCR-SSP), HLA-DR phenotype and genotype frequencies were ascertained in 86 RA hospital out-patients and 79 healthy controls matched for age, gender and ethnicity. HLA-DR1 and HLA-DR4 frequency was similar in patients and controls. HLA-DR10 occurred in 26 instances (15%) in RA and in eight (5%) controls (Pcorr = 0.048). HLA-DR2 was also increased in patients (P = 0.053) and its major subtype DR15 was significantly increased (Pcorr = 0.03). HLA-DR5 frequency was 5% in patients and 19% in controls (Pcorr = 0.002). The HLA-DR4 alleles possessing the shared epitope were more common in RA (Pcorr = 0.03) and this difference was enhanced by inclusion of other alleles possessing the shared epitope (Pcorr = 0.002). Shared epitope alleles were observed in 43 (50%) patients and 17 (22%) controls (Pcorr = 0.003). The shared epitope did not distinguish patients with more severe disease, as reflected by pain, joint deformities, disability, rheumatoid factor or X-ray damage. The distribution of HLA-DR alleles in Pakistanis with RA supports the shared epitope hypothesis. In common with other non-European racial groups, HLA-DR4 was not associated with RA. Unlike other groups, there was a weak link of RA with HLA-DR2. A protective effect of HLA-DR5 was apparent. In accord with some other studies, the shared epitope in this hospital out-patient population was not a marker for more severe disease.