We aimed to evaluate the value of magnetic resonance imaging (MRI) for the diagnosis of intra-articular adhesions of the temporomandibular joint (TMJ). We diagnosed 27 consecutive patients with internal derangement of the TMJ (33 TMJs) between 1 December and 31 July 2003 in our department. All patients were examined by MRI and arthroscopy. The MRI findings were recorded as "positive", "suspicious", or "negative", and were compared with those of arthroscopy to obtain the numbers of true positive, true negative, false positive, and false negative results. Arthroscopy confirmed that 18 of the 33 TMJs contained adhesions. MRI showed that 11 were "positive", 4 were "suspicious", and 18 were "negative". After comparing these results with those of arthroscopy, 7 TMJs were true positives and 4 were false positives of the 11 "positive" TMJs. Three of the 4 "suspicious" TMJs had adhesions. Of the 18 negative TMJs, 10 were true negatives and 8 were false negatives. We concluded that the diagnostic accuracy of MRI for intra-articular adhesions was poor; most of the adhesions were not diagnosed by MRI, but intracapsular adhesions could be detected on T2 weighted-images with existing synovial fluid.