[Objective] To investigate factors related to failure of arthroscopic rotator cuff repair with anchor suture and the outcomes of corresponding revision arthroscopies. [Methods] From June 2012 to June 2020, a total of 18 patients, including 8 males and 10 females aged 45~73 years with a mean of (58.2±4.5) years, received corresponding revision arthroscopies for anchor failure in primary arthroscopic rotator cuff repair for its tear. [Results] All the 18 patients were proved subacromial synovial hyperplasia and partial rotator cuff re-tear un- der arthroscopic examination. Among them, 5 cases were proved of anchor pull out, accounting for 27.8%; 9 cases were proved of anchor loosening, accounting for 50.0%; and 4 cases of suture tied loosening, accounting for 22.2%; which was statistically significant (P＜0.05) . All the 18 patients underwent revision surgery with operation time of 45~110 min, with an average of (63.7±11.3) min, whereas without seri- ous complications such as vascular and nerve injuries. After more than 12 months of follow- up, the VAS, Neer and Constant- Murley scores, as well as forward flexion-lifting, abduction-lifting and lateral external rotation range of motions (ROMs) of the affected side were significantly improved compared with those before revision surgery (P＜0.05) . The affected side was significantly inferior to the healthy side in terms of abovementioned parameters preoperatively (P＜0.05) , whereas which became not statistically significant between the two sides at the latest follow-up (P＞0.05) . Radiographically, all the patients had anchors remained in good position after operation without loosening or displacement until the latest follow-up. Compared with those preoperatively, the MRI rotator cuff grade on the affected side improved (P＜ 0.05) , the acromiohumeral interval (AHI) remained unchanged at the latest follow up (P＞0.05) . The affected side proved significantly infe- rior to the healthy side in MRI rotator cuff grade preoperatively (P＜0.05) , but there was no significant difference in AHI between the two sides (P＞0.05) . At last follow-up, there was no significant difference in MRI rotator cuff grade and AHI between the two sides (P＞0.05) . [Conclusion] Anchor failure may occur in the arthroscopic repair of rotator cuff injury, anchor loosening is the most common one among them. According to the specific situation, removing or retaining the loosed anchor and replacing the new anchors to repair the rotator cuff can still obtain satisfactory clinical results.