We conducted a prospective, randomized trial to compare the safety and effectiveness of the A-V Impulse System foot pump with that of low-molecular-weight heparin for reducing the prevalence of deep-vein thrombosis after total hip replacement. Of 290 patients who were to have a primary total hip replacement, 143 were randomized to receive enoxaparin (forty milligrams daily) for seven days after the operation and 147, to use the foot pump for seven days. The primary outcome measure was the prevalence of deep-vein thrombosis, as determined by venography on the sixth, seventh, or eighth postoperative day. Secondary outcome measures included transfusion requirements, intraoperative blood loss, postoperative drainage, blood-loss index, appearance of the site of the wound according to a subjective visual-analog scale, and swelling of the thigh. The patients' compliance with the regimen for use of the foot pump was monitored with an internal timing device, and their acceptance of the device was assessed with a questionnaire. Symptoms consistent with pulmonary embolism were investigated with ventilation-perfusion scanning. The patients were contacted later for detection of symptoms of venous thromboembolism that may have occurred during the first three months after discharge from the hospital.Venography was performed on 274 patients: 136 who used the foot pump and 138 who received enoxaparin. Deep-vein thrombosis was detected in twenty-four (18 per cent) of the patients who used the foot pump compared with eighteen patients (13 per cent) who received enoxaparin (95 per cent confidence interval for the difference in proportions, -3.9 to +13.0 per cent). Thrombosis in the calf was found in seven patients (5 per cent) in the former group compared with six patients (4 per cent) in the latter (95 per cent confidence interval for the difference, -4.2 to +5.8 per cent), and proximal thrombosis was observed in seventeen patients (13 per cent) in the former group compared with twelve patients (9 per cent) in the latter (95 per cent confidence interval for the difference, -3.5 to +11.1 per cent). None of these differences was significant. No patient in either group had major proximal deep-vein thrombosis; all proximal thrombi were isolated entities involving the femoral valve cusp and were of unknown importance. One patient who used the foot pump had a non-fatal pulmonary embolism. One patient who received enoxaparin had a symptomatic deep-vein thrombosis during hospitalization. Two patients (one from each group [0.7 per cent]) were readmitted to the hospital because of a symptomatic deep-vein thrombosis despite normal venographic findings at the time of discharge. There was no difference in the transfusion requirements or the intraoperative blood loss between the two groups.There were more soft-tissue side effects in the patients who received enoxaparin than in those who used the foot pump: there was more bruising of the thigh and oozing of the wound (p < 0.001 for each), postoperative drainage (578 compared with 492 milliliters; p = 0.014), and swelling of the thigh (twenty compared with ten millimeters; p = 0.03). Of 124 patients who used the foot pump and were asked about the acceptability of the device, fourteen (11 per cent) said that it was uncomfortable, twenty-one (17 per cent) reported sleep disturbance, and four (3 per cent) stated that they had stopped using the device. Conversely, ten (8 per cent) found it relaxing.We concluded that the foot pump is a suitable alternative to low-molecular-weight heparin for prophylaxis against thromboembolism after total hip replacement and that it produces fewer soft-tissue side effects. Tolerance of the device is a problem for some patients.