We report a randomised controlled trial to examine the effectiveness and cost-effectiveness of arthroscopic acromioplasty in the treatment of stage II shoulder impingement syndrome. A total of 140 patients were randomly divided into two treatment groups: supervised exercise programme (n = 70, exercise group) and arthroscopic acromioplasty followed by a similar exercise programme (n = 70, combined treatment group). The main outcome measure was self-reported pain on a visual analogue scale of 0 to 10 at 24 months, measured on the 134 patients (66 in the exercise group and 68 in the combined treatment group) for whom endpoint data were available. An intention-to-treat analysis disclosed an improvement in both groups but without statistically significant difference in outcome between the groups (p = 0.65). The combined treatment was considerably more costly. Arthroscopic acromioplasty provides no clinically important effects over a structured and supervised exercise programme alone in terms of subjective outcome or cost-effectiveness when measured at 24 months. Structured exercise treatment should be the basis for treatment of shoulder impingement syndrome, with operative treatment offered judiciously until its true merit is proven.