The authors undertook a randomised prospective study to investigate the contribution of thigh tourniquets to the formation of intra-operative venous emboli during lower limb surgery. Patients were randomised to have a thigh tourniquet or no tourniquet and transoesophageal echocardiography was used to detect embolic signals in the right heart during and after knee arthroscopy. Three physicians blinded to patient demographics and tourniquet status separately assessed videotapes of the echocardiograms for evidence of emboli. Of the 32 patients randomised, 18 underwent knee arthroscopy with and 14 without tourniquet. Emboli were seen in 72% (95% CI 55 to 84) of patients, in 14 patients with tourniquet and in 9 patients without tourniquet. There was an estimated 13% greater incidence of emboli in the tourniquet group compared to the non-tourniquet group, a difference which was not statistically significant (Fisher's Exact Test, p = 0.45). No patients suffered symptoms or signs attributable to a pulmonary embolus.