A total of 159 patients (84 women and 75 men, mean age of 53 (20 to 87)) with subacromial impingement were randomised to treatment with subacromial injections using lidocaine with one of hyaluronic acid (51 patients), corticosteroid (53 patients) or placebo (55 patients). Patients were followed up for 26 weeks. The primary outcome was pain on a visual analogue score (VAS), and secondary outcomes included the Constant Murley score, shoulder pain score, functional mobility score, shoulder disability questionnaire and pain-specific disability score. The different outcome measures showed similar results. After three, six and 12 weeks corticosteroid injections were superior to hyaluronic acid injections and only at six weeks significantly better than placebo injections. The mean short-term reduction in pain on the VAS score at 12 weeks was 7% (SD 2.7; 97.5% confidence interval (CI) 0.207 to 1.55; p = 0.084) in the hyaluronic acid group, 28% (SD 2.8; 97.5% CI 1.86 to 3.65; p <0.001) in the corticosteroid group and 23% (SD 3.23; 97.5% CI 1.25 to 3.26; p <0.001) in the placebo group. At 26 weeks there was a reduction in pain in 63% (32 of 51) of patients in the hyaluronic acid group, 72% (38 of 53) of those in the corticosteroid group and 69% (38 of 55) of those in the placebo group. We were not able to show a convincing benefit from hyaluronic acid injections compared with corticosteroid or placebo injections. Corticosteroid injections produced a significant reduction in pain in the short term (three to 12 weeks), but in the long term the placebo injection produced the best results.