We assessed the influence of the use of a tourniquet in total knee arthroplasty in a prospective, randomised study. After satisfying exclusion criteria, we divided 77 patients into two groups, one to undergo surgery with a tourniquet and one without. Both groups were well matched. The mean change in knee flexion in the group that had surgery without a tourniquet was significantly better at one week (p = 0.03) than in the other group, but movement was similar at six weeks and at four months. There was no significant difference in the surgical time, postoperative pain, need for analgesia, the volume collected in the drains, postoperative swelling, and the incidence of wound complications or of deep-venous thrombosis. We conclude that the use of a tourniquet is safe and that current practice can be continued.