BACKGROUND Shoulder maneuvers and magnetic resonance imaging (MRI) are performed to diagnose supraspinatus tendon tears regardless of arthroscopy exam. Although there are many studies on this subject, there is a lack of studies comparing the sensitivity (Se) and specificity (Sp) of shoulder maneuvers and MRI to arthroscopic findings (intact, partial, or full thickness supraspinatus tendon tear). AIM To compare the diagnostic values of shoulder maneuvers with MRI for supraspinatus tendon tears in patients undergoing shoulder arthroscopy. METHODS A total of 199 consecutive patients from four orthopedic centers met the eligibility criteria of shoulder pain persisting for at least four weeks. They were prospectively enrolled in this study from April 2017 to April 2019. Seven clinical tests (full can, empty can, drop arm, Hawkins’, painful arc, Neer’s sign and resisted external rotation) and MRI were performed, and all were compared with surgical findings. Full can, empty can and resisted external rotation tests were interpreted as positive in the case of pain and/or weakness. We assessed the Se, Sp, accuracy, positive predictive value (PPV) and negative predictive value (NPV), positive and negative likelihood ratio and diagnostic odds ratio for overall, partial and full-thickness supraspinatus tears. RESULTS MRI had the highest Se for overall (0.97), partial (0.91) and full-thickness (0.99) tears; moreover, MRI had the highest NPV: 0.90, 0.88 and 0.98 for overall, partial and full-thickness tears, respectively. For overall supraspinatus tears, the Se and PPV were: Painful arc (Se = 0.85/PPV = 0.91), empty can (pain) (Se = 0.80/PPV = 0.89), full can (pain) (Se = 0.78/PPV = 0.90), resisted external rotation (pain) (Se = 0.48/PPV = 0.87), drop arm (Se = 0.19/PPV = 0.97), Neer’s sign (Se = 0.78/PPV = 0.93) and Hawkins’ (Se = 0.80/PPV = 0.88). MRI had the highest PPV (0.99). The Hawkin’s test had the highest false positive rate in patients with intact tendons (0.36). The Sp of the empty can and full can (both tests positive for pain and weakness), drop arm and MRI were: 0.93, 0.91, 0.98 and 0.96, respectively. For partial and full-thickness tears, the empty can test (positive for pain and weakness) had a Sp of 0.93, and the drop arm and MRI had the same Sp (0.98). CONCLUSION Physical examination demonstrated good diagnostic value, the drop arm test had a Sp as good as MRI for supraspinatus tears; however, MRI was more accurate in ruling out tears. The Hawkins’ test had high false-positive findings in patients with intact tendons.