We report a prospective study of 49 patients who had arthroscopic subacromial decompression for chronic rotator-cuff impingement. All patients were assessed preoperatively and at 3, 6 and 12 months using the modified UCLA shoulder score. The dominant arm was affected in 35 patients, but only 13 recognised overuse as a cause of their shoulder pain. Before operation, the UCLA shoulder score was poor or fair in all patients. After three months only 28% of patients had satisfactory