Purpose To evaluate the diagnostic accuracy between magnetic resonance arthrography (MRA) and arthroscopic examination for the assessment of pathologies of the long head of the biceps (LHB) prior to rotator cuf (RC) repair. The hypothesis was that MRA is suitable to identify biceps instabilities, due to improved visibility of the biceps pulley. Methods Sixty-six patients aged 58.5 ± 17.6 (range, 46-71) scheduled to have RC repair between 2016 and 2017 were prospectively enrolled. MRA images of the LHB were interpreted by one radiologist and two surgeons, then compared to arthroscopic indings. The sensitivity, speciicity, positive predictive value (PPV) and negative predictive value (NPV) were calculated by comparing MRA axial readings to arthroscopic assessment in the neutral position (static instability), MRA sagittal readings to arthroscopic assessment in external rotation (dynamic s instability), MRA coronal and axial readings to arthroscopic assessment in the neutral position (tendinopathy). Results Among the three observers, the radiologist obtained the highest sensitivity and speciicity for MRA, which were respectively: ( 1) for static instability, 62% (C.I. 35-85) and 77% (C.I. 63-88); (2) for dynamic instability, 50% (C.I. 29-71) and 62% (C.I. 46-77), and (3) for tendinopathy, 49% (C.I. 36-62) and 100% (C.I. 3-100). Conclusions MRA is not suitable for the diagnosis of LHB lesions prior to arthroscopic rotator cuf repair. Level of evidence Diagnostic study, Level I. Keywords Magnetic resonance arthrography • Arthroscopic rotator cuf repair • Lesion of the long head of the biceps • Diagnostic accuracy