OBJECTIVE A prospective study to assess the value of indirect magnetic resonance-arthrography (MR-Arthro) in the detection of lesions of the superior labrum (SLAP-lesions). METHODS Unenhanced magnetic resonance imaging (MRI) and indirect MR-Arthro of the shoulder were performed in 35 patients (9 females and 26 males) with a history of acute or chronic shoulder disorder. The images of unenhanced MRI and indirect MR-Arthro were analyzed independently from each other for the occurrence of SLAP-lesions and the findings correlated with the arthroscopic results. For arthroscopy, MRI and MR-Arthro, the SLAP-lesions were categorized based on the classification system of Snyder. RESULTS SLAP-lesions were diagnosed by arthroscopy in 22 of 35 patients (63 %). 9 of the 22 lesions (41 %) were classified as type 1, 9 (41 %) as type 2, three (13.5 %) as type 3 and one (4.5 %) as type 4. Using unenhanced MRI sensitivity, specificity and accuracy in the detection of SLAP-lesions were calculated to be 73 %, 85 % and 77 %. Indirect MR-Arthro showed a sensitivity of 91 %, a specificity of 85 % and an accuracy of 89 %. Unenhanced MRI revealed correlation to the arthroscopic classification in 9 of 22 cases (41 %) and indirect MR-Arthro in 18 of 22 cases (82 %). CONCLUSIONS Indirect MR-Arthro is a non-invasive method which offers excellent sensitivity in the diagnosis of SLAP-lesions. It provides important preoperative informations with regard to the exact location and extent of a tear.