PURPOSE: Tendon tear may result in muscular retraction with the loss of contractile amplitude and strength of the rotator cuff muscles. Currently, neither a validated method of measuring supraspinatus tendon length nor normal values are known. It was therefore the purpose of this study to measure the normal length of the supraspinatus tendon and to determine whether partial tears are associated with changes in tendon length. METHODS: MR examinations of 49 asymptomatic volunteers and 37 patients with arthroscopically proven, isolated partial tears of the supraspinatus tendon were compared. The ratio of the extramuscular tendon length to the distance between the footprint and the glenoid surface was calculated (TL/FG ratio). Tendon length measurements were taken by two independent readers at the bursal and articular surfaces at the anterior, the central and the posterior parts of the tendon. RESULTS: TL/FG ratios at the bursal surface of tendons with partial tears were significantly higher than those in the control group [anterior: 0.78 ± 0.20 vs. 0.66 ± 0.15 (p < 0.05); central: 0.61 ± 0.13 vs. 0.52 ± 0.10 (p < 0.05); posterior: 0.57 ± 0.15 vs. 0.52 ± 0.10 (p < 0.05)]. At the articular surface, differences were significant only anteriorly [0.60 ± 0.13, vs. 0.54 ± 0.10 (p < 0.05)]. A cut-off TL/FG ratio of 0.63 for measurements at the bursal surface in the center of the tendon achieved a sensitivity of 46 % and a specificity of 92 % for the identification of partial cuff tearing. CONCLUSION: A reproducible method for measurement of extramuscular supraspinatus tendon length is described. Partial tearing of the supraspinatus tendon is associated with significant tendon lengthening, suggesting failure in continuity, and this is most reliably measured on the bursal surface. LEVEL OF EVIDENCE: III.