Introduction and purpose: Since 1834 when Smith first described rotator cuff tears we have come far in the treatment of this condition. The greater availability of arthroscopic surgical procedures requires the greatest possible precision in preoperative diagnoses. Our purpose is to determine the accuracy (sensitivity and specificity), in our environment, of the two main diagnostic techniques we use, namely ultrasound and magnetic resonance.

Materials and methods: We reviewed 100 arthroscopic shoulder procedures performed in our centre. We compared retrospectively the intraoperative findings with the reports of the available images of each patient (US and MRI). We focused our attention exclusively on the tendon of the supraspinatus muscle. The sensitivity and specificity of both tests was compared in two different ways. In the first case, by considering as torn tendons those with complete tears (total thickness tears); in the second case, by considering as torn tendons those with complete or partial tears.

Results: With magnetic resonance images we found a sensitivity of 78% and a specificity of 89% for total tears and a sensitivity of 92% and a specificity of 64% for partial tears. With ultrasound we found a sensitivity of 73% and a specificity of 83% for total tears and a sensitivity of 91% and a specificity of 78% for partial tears.

Conclusions: Both diagnostic methods show comparable sensitivity and specificity levels, which were high in our environment for total tears. Specificity is less in the case of partial tears.