Evidence-based clinical diagnostic criteria for temporomandibular joint (TMJ) arthritis are not available. To establish (i) criteria for clinical diagnosis of TMJ arthritis and (ii) clinical variables useful to determine inflammatory activity in TMJ arthritis using synovial fluid levels of inflammatory mediators as the reference standard. A calibrated examiner assessed TMJ pain, function, noise and occlusal changes in 219 TMJs (141 patients, 15 healthy individuals). TMJ synovial fluid samples were obtained with a push-pull technique using the hydroxycobalamin method and analysed for TNF, TNFsRII, IL-1, IL-1ra, IL-1sRII, IL-6 and serotonin. If any inflammatory mediator concentration exceeded normal, the TMJ was considered as arthritic. In the patient group, 71% of the joints were arthritic. Of those, 93% were painful. About 66% of the non-arthritic TMJs were painful to some degree. Intensity of TMJ resting pain and TMJ maximum opening pain, number of jaw movements causing TMJ pain and laterotrusive movement to the contralateral side significantly explained presence of arthritis (AUC 0.72, P<.001). Based on these findings, criteria for possible, probable and definite TMJ arthritis were determined. Arthritic TMJs with high inflammatory activity showed higher pain intensity on maximum mouth opening (P<.001) and higher number of painful mandibular movements (P=.004) than TMJs with low inflammatory activity. The combination TMJ pain on maximum mouth opening and Contralateral laterotrusion <8mm appears to have diagnostic value for TMJ arthritis. Among arthritic TMJs, higher TMJ pain intensity on maximum mouth opening and number of mandibular movements causing TMJ pain indicates higher inflammatory activity.