The authors studied the use of epidural anesthesia prolonged to 48 hours of epidural analgesia in 120 total hip arthroplasty patients in a case-control fashion. One half of the patients received prolonged epidural anesthesia, while the other matched half received general endotracheal anesthesia. Venograms were obtained after surgery and graded in blind fashion by a single radiologist. The overall incidence of deep venous thrombosis in the epidural versus general anesthetic groups was 23 (14 of 60 patients) versus 40% (24 of 60) (P < .05). There was an identical incidence--8.3% (5 of 60 patients)--of proximal thrombosis in the two groups, and all of the difference in the overall rates of thrombosis occurred in the calf. Fifteen percent (9 of 60 patients) of the epidural patients and 31.6% (19 of 60) of the general anesthetic patients demonstrated this finding (P < .05). Of the 10 proximal clots, 8 (80%) were found in the operative leg, while only 29 (59.2%) of the 49 calf clots were found in the operative leg. Prolonged epidural anesthesia significantly decreases the incidence of deep venous thrombosis after total hip arthroplasty, with its most apparent benefit on calf vein thrombosis secondary to its hyperkinetic effect on lower limb blood flow. The observation that it has no demonstrable effect on the prevention of proximal thrombosis and our finding that the majority of proximal clots are in the operative leg suggest that thrombi in the thigh may be the result of a different primary pathogenic mechanism that is more related to endothelial injury than to changes in viscosity or blood flow.(ABSTRACT TRUNCATED AT 250 WORDS)