PURPOSE: To describe and evaluate a simple perioperative test (the derotation sign) that differentiates significant (grade 3) partial-thickness and small full-thickness rotator cuff tears from insignificant (grades 1 and 2) partial-thickness rotator cuff tears and intact rotator cuffs. TYPE OF STUDY: Sensitivity and specificity study. METHODS: A study was conducted of 123 patients who underwent shoulder arthroscopy for chronic symptomatic rotator cuff disease, each of whom was subjected to a derotation test under general anesthesia. Specifically, the glenohumeral joint was rapidly distended with fluid just before arthroscopy. One of 3 possible observations was made: (1) the arm rotated internally only, (2) the arm initially rotated internally and then externally rotated (positive derotation sign), or (3) the arm did not rotate at all. Arthroscopic findings were then correlated with the derotation test. RESULTS: Forty-one shoulders had no arm rotation on derotation testing; all had rotator cuff tears greater than 2 cm in size. Forty-two shoulders with impingement and no rotator cuff tears, as well as 23 shoulders with grade 1 or grade 2 partial-thickness rotator cuff tears showed internal rotation only. Seventeen shoulders had a positive derotation sign; all of these had either a grade 3 partial-thickness or a small full-thickness (<1 cm) rotator cuff tear. CONCLUSIONS: The derotation test is useful in the perioperative differentiation of functionally intact rotator cuffs from those with significant tears. The derotation sign is specific for the diagnosis of grade 3 partial-thickness and/or small (<1 cm) full-thickness rotator cuff tears.