A total of 92 patients with symptoms for over six months due to subacromial impingement of the shoulder, who were being treated with physiotherapy, were included in this study. While continuing with physiotherapy they waited a further six months for surgery. They were divided into three groups based on the following four clinical and radiological criteria: temporary benefit following steroid injection, pain in the mid-arc of abduction, a consistently positive Hawkins test and radiological evidence of impingement. Group A fulfilled all four criteria, group B three criteria and group C two criteria. A total of nine patients improved while waiting for surgery and were excluded, leaving 83 who underwent arthroscopic subacromial decompression (SAD). The new Oxford shoulder score was recorded pre-operatively and at three and 12 months post-operatively. A total of 51 patients (group A) had a significant improvement in the mean shoulder score from 18 (13 to 22) pre-operatively to 38 (35 to 42) at three months (p < 0.001). The mean score in this group was significantly better than in group B (21 patients) and C (11 patients) at this time. At one year patients in all groups showed improvement in scores, but patients in group A had a higher mean score (p = 0.01). At one year patients in groups A and B did better than those in group C (p = 0.01). Arthroscopic SAD is a beneficial intervention in selected patients. The four criteria could help identify patients in whom it is likely to be most effective.