As rheumatoid arthritis (RA) is an HLA-DR associated autoimmune disease and soluble HLA-DR (sHLA-DR) molecules have the capacity to regulate the immune response, we studied the sHLA-DR levels in RA patients in view of therapy modalities and clinical and biologic parameters of disease activity. For this sHLA-DR concentrations from 87 RA patients were determined by a sensitive enzyme-linked immunoabsorbent assay (ELISA) format. There was a weak but significant correlation between sHLA-DR levels and disease activity (r 0.186 to 0.287, p <0.004 to <0.001). The mean serum sHLA were not significantly different between groups with or without corticosteroids, or undergoing therapy with different disease modifying antirheumatic drugs. However, patients treated with a combination of methotrexate and prednisolone have lower sHLA-DR (206 ± 21 ng/ml, n = 34) compared with the mean value for all other samples (306 ± 16, n = 217, p <0.001). This corresponded with significantly lower EULAR pain and swelling scores, ESR and rheumatoid factor (RF) by latex fixation (p <0.02 to 0.001) in the former, compared with the latter group. Furthermore, sHLA-DR was, respectively, 267 ± 15 ng/ml (n = 182) in samples from patients treated with nonsteroidal anti-inflammatory drugs (NSAIDs), and 358 ± 31 (n = 72) without NSAIDs (p <0.01). Lower sHLA-DR with NSAIDs contrasted with significantly higher scores for pain, swelling, CRP, and RF by latex fixation and by Waaler-Rose test (p <0.05 to 0.001). Comparison of subgroups with or without the shared epitope of RA disease (Q)R/KRAA within the HLA-DR _1-chain confirmed significantly higher parameters of disease activity and sHLA-DR in the presence of this disease associated epitope in our patients. Different mechanisms appear to be involved in sHLA-DR production or release, as their level correlates positively with disease activity under combined therapy with corticosteroids and methotrexate, but decreases with higher disease activity in patients treated with NSAIDs