Ultrasonography (US) is being increasingly used in clinical practice to detect rotator cuff tears (RCTs) in patients with rheumatoid arthritis (RA) who have shoulder pain. The major aim of this study was to determine the diagnostic utility of US and magnetic resonance imaging (MRI) for detecting RCTs in patients with RA who have persistent shoulder pain. Patients and methods: With standardized procedures, US and MRI examinations of the shoulder were performed in 36 patients with RA who had persistent shoulder pain prior to arthroscopic intervention. Within 1 month after US and MRI examination, arthroscopic repair was performed. Arthroscopic findings were used as the gold standard for the diagnosis of RCTs. Results: Full-thickness RCTs in 28 patients with RA (77.8%) and partial-thickness RCTs in eight patients (22.2%) were identified using arthroscopic inspection. With arthroscopic findings as the gold standard, the sensitivity and accuracy of US in detecting full-thickness RCTs were 92.9% and 89%, respectively, whereas those for MRI were 96.4% and 90%, respectively. In detecting partial-thickness RCTs, the sensitivity and accuracy were 62.5% and 75.0%, respectively, for US, in contrast with 87.5% and 88%, respectively, for MRI. The overall agreement between US and MRI was 89.3% in detecting full-thickness RCTs and 75.0% in detecting partial-thickness RCTs. US demonstrated levels of sensitivity similar to that of MRI in detecting posterior recess synovitis, tenosynovitis, and subacromial-subdeltoid bursitis. Conclusion: With a good agreement with MRI, US was shown to be a highly sensitive and accurate imaging modality in detecting full-thickness RCTs for patients with RA who have shoulder pain, but appeared to have lower sensitivity in detecting partial-thickness RCTs compared with MRI.