Purpose of the study The authors examined the anatomic condition and the function of the rotator cuff obtained after an average period of four years following surgical repair in a series of 100 full thickness rotator cuff tears. The aim was to assess the validity of Constant's scoring method and to analyse risk factors and the frequency of recurrent tears. Material and methods The series comprised 98 patients, 62 men and 36 women whose average age was 56 years. It included 69 tears of less than 2 cm in size (39 cases) or between 2 to 4 cm (3 cases) of the supra-spinatus, 22 tears of the supra-and infraspinatus measuring between 2 to 4 cm, and 9 massive tears. The tendon of the long head of the biceps was pathological in 1/3 of cases. All 98 patients were operated on by the same surgeon using the same repair technique, and all followed ambulatory rehabilitation along the same principles of self-rehabilitation applied pre operatively. In each patient function was assessed using Constant's scoring method, and the condition of the repaired cuff was determined by ultrasonography at the time of clinical follow-up. The average follow-up period was 4 years (2 to 6 years). Results Ultrasonography revealed intact cuffs in 65 per cent, thinned cuffs in 11 per cent and recurrent full thickness tears in 24 per cent of cases. The risk of recurrent tear increased with the extent of the tear to be repaired (57 per cent), in older patient (25 per cent) and with a higher lever of post-surgical occupational use (18 per cent). A drop in the post-operative Constant score had a predictive value for a full thickness recurrent defect. Discussion Constant's scoring method appears to be a reliable, reproducible method for analysing functional results following surgical repair of full thickness cuff tears and to reflect the anatomic condition of the repaired cuff. At clinical follow-up, the anatomic condition of the cuff is more determinant of final functional results than initial tear size. Conclusions Assessment of functional results must be complemented by anatomic examination using ultrasonography in order to specify the size of any possible recurrent defect and to detect thinning of the cuff which cannot be identified by Constant's score. Analysis of the risk factors for recurrent tear led the authors to question the necessity of repairing massive tears in older patients and pointed to the valuable advantages of reinforcing fragile cuffs during initial repair especially in very active patients.