Abstract: MRI is the primary examination tool for assessing rotator cuff tears, which could not only intuitively display the rotator cuff tear area (such as the size, location, and degree of fat infiltration of the tear）, but it can also qualitatively and quantitatively analyze each tendon and muscle. The application of deep learning (DL) in MRI diagnosis of rotator cuff tears is a new field of artificial intelligence research, which provides a wide application prospects for musculoskeletal radiology, such as improving work efficiency and diagnostic accuracy. The authors summarized the research progress of DL in MRI diagnosis of rotator cuff tears in the last ten years to provide new ideas for the diagnosis, treatment, prognosis evaluation, and follow-up of rotator cuff tears.