The extent of fatty infiltration of rotator cuff muscles was evaluated on computed tomography displays using a five-point scoring system in 63 shoulders before surgery for a rotator cuff tear and in 57 of these shoulders after surgical repair of the tear (the quality of cuff repair was evaluated in these 57 shoulders by arthrography, usually coupled with computed tomography, after a mean time interval of 8 months since the procedure). Preoperatively, significant fatty infiltration of the supraspinatus muscle was uncommon. Severe fatty infiltration of the infraspinatus and subscapularis muscles was seen in some patients. In most instances, fat accumulated in those muscles whose tendons were torn; however, tendon cleavage was associated with mild fatty infiltration of the corresponding muscle in some instances, and wide tears of the supraspinatus and subscapularis muscles were sometimes accompanied with severe fatty infiltration of the infraspinatus muscle. In half the cases, fatty degenerescence of the infraspinatus muscle occurred within six months of the tendon rupture. After surgery, in most shoulders without evidence of rotator cuff leakage on the postoperative arthrogram, no further accumulation of fat occurred. However, a decrease in the amount of existing fat was rarely documented and occurred only in the supraspinatus muscle. Before and after surgery, increasing severity of the fatty infiltration of the infraspinatus muscle was associated with increasing functional impairment of the same muscle. Recurrent rotator cuff tears, which involved only the supraspinatus muscle, were considerably more common when there was severe preoperative fatty infiltration of the infraspinatus muscle.(ABSTRACT TRUNCATED AT 250 WORDS)