To compare three different ultrasound-guided injections for chronic tennis elbow.Assessor-blinded, randomized controlled comparative trial.44 patients with clinically diagnosed tennis elbow, confirmed by Doppler ultrasound, received under ultrasound guidance, a single corticosteroid injection (n=14), or two injections (separated by 4 weeks) of either autologous blood (n=14) or polidocanol (n=16). Clinical and ultrasound examination was performed at baseline, 4, 12 and 26 weeks.Complete recovery or much improvement was greater for corticosteroid injection than autologous blood and polidocanol at 4 weeks (p<0.001, number needed to treat 1 (95% CI 1–2)). In contrast, at 26 weeks corticosteroid was significantly worse than polidocanol (p=0.004, number needed to harm 2 (1–6)). Recurrence after corticosteroid injection was significantly higher than autologous blood or polidocanol (p=0.007, number needed to harm 2 (1–4)). Corticosteroid injection produced greater reduction in tendon thickness and vascularity than autologous blood at 4 weeks only. Compared to autologous blood, polidocanol reduced tendon thickness at 4 and 12 weeks and reduced echogenicity and hyperaemia after 12 or 26 weeks respectively.Injections of corticosteroid cannot be recommended over polidocanol or autologous blood, because despite beneficial short-term effect there were inferior long-term effects. Whether polidocanol or autologous blood injections are effective is unknown, especially as their global effect profiles are not unlike previously reported for wait-and-see.