Epidural anesthesia has been reported to reduce the prevalence of deep-vein thrombosis after total hip arthroplasty compared with the prevalence after general anesthesia. However, the effect of epidural anesthesia on the rate of thrombosis after total knee arthroplasty has not been reported previously, to our knowledge. A review was conducted of 705 total knee arthroplasties (541 patients) that had been performed by a single surgeon between September 1984 and December 1988. During this period, the operative technique, the protocol for rehabilitation, and the regimen for prophylaxis against thromboembolism did not change meaningfully. The patients received either epidural or general anesthesia. Preoperative and postoperative perfusion scans of the lungs and a venogram of the lower limb or limbs that had been operated on were done for all patients. For the 227 patients who had received epidural anesthesia, the over-all rate of deep-vein thrombosis was 48 per cent, which was significantly lower than the 64 per cent incidence in the 264 patients who had received general anesthesia (p less than 0.0001). The greatest reduction was in the occurrence of proximal thrombosis, which was identified in 9 per cent of the patients who had had general anesthesia but in only 4 per cent of those who had had epidural anesthesia (p less than 0.05). The use of epidural anesthesia reduced the incidence of proximal thrombosis after both unilateral and one-stage bilateral arthroplasty.