We present the results of arthroscopic repair of full-thickness tears of the rotator cuff in seventy-three patients (thirty-nine men and thirty-four women). The average age of the patients at the time of the operation was 60.7 years (range, thirty-one to eighty-two years). All of the patients were followed for at least two years (average, thirty months; range, twenty-four to forty months). The shoulders were evaluated with the rating scale of the University of California at Los Angeles, the shoulder index of the American Shoulder and Elbow Surgeons, and the functional rating scale of Constant and Murley. In addition, the patients completed the Short-Form 36 Health Survey (SF-36) preoperatively and at the yearly follow-up evaluations. Eleven tears were small (less than one centimeter in length), forty-five were medium (one to three centimeters), eleven were large (more than three to five centimeters), and six were massive (more than five centimeters). The average length of the tear was twelve millimeters, and the average width was twenty-seven millimeters. Sixty-nine tendons were repaired anatomically, and four were repaired an average of three millimeters (range, two to eight millimeters) medial to the anatomical insertion of the tendon. An average of 2.3 (range, one to four) suture anchors were used in the repair. Sixty-three glenohumeral joints were normal, and ten had an intra-articular lesion. Seven patients had a concomitant resection of the acromioclavicular joint. The average duration of the operation was fifty-six minutes (range, thirty-five to ninety minutes). The active and passive ranges of motion improved significantly after the procedure (p = 0.0001). The strength of resisted elevation improved from 7.5 to 14.0 pounds (3.4 to 6.3 kilograms) (p = 0.0001). The average total score according to the rating scale of the University of California at Los Angeles improved from 12.4 to 31.1 points; the average total score according to the shoulder index of the American Shoulder and Elbow Surgeons, from 30.7 to 87.6 points; and the average absolute score according to the rating system of Constant and Murley, from 41.7 to 83.6 points (p = 0.0001 for all comparisons). The average score for the pain component of the rating scale of the University of California at Los Angeles improved from 2.4 to 8.6 points; fifty-seven (78 per cent) of the seventy-three patients rated the relief of pain as good or excellent on the visual-analog scale. The average score for satisfaction improved from 0.4 to 4.6 points; sixty-six patients (90 per cent) rated their satisfaction as good or excellent at the time of the most recent examination. None of the shoulders were rated as good or excellent before the operation, whereas sixty-one (84 per cent) were so rated at the most recent follow-up evaluation after the index procedure. In addition, significant improvements (p = 0.0015) were noted in the scales and summary measures of the SF-36. Arthroscopic repair of full-thickness tears of the rotator cuff produced satisfactory results with regard to traditional orthopaedic criteria as well as with regard to patient-assessed criteria such as satisfaction, pain relief, and general health. The arthroscopic method offers several advantages, including smaller incisions, access to the glenohumeral joint for the inspection and treatment of intra-articular lesions, no need for detachment of the deltoid, and less soft-tissue dissection. However, these advantages must be considered against the technical difficulty of the method, which limits its application to surgeons who are skilled in both open and arthroscopic procedures on the shoulder.