Duplex ultrasonography was used to screen 273 consecutive patients, on the fourth day after a total hip or total knee arthroplasty, for the presence of a distal deep venous thrombosis. Patients who had a history of thromboembolic disease or who had an active neoplasm were excluded from the study. Of the 273 patients, twenty-four (9 per cent) were found to have a proximal deep venous thrombosis (a thrombosis involving the femoral or popliteal veins) and forty-one (15 per cent), a distal deep venous thrombosis (a thrombosis involving the veins of the calf). All of the thromboses were asymptomatic. The forty-one patients in whom a distal deep venous thrombosis had been detected with duplex screening subsequently had serial duplex examinations, on the seventh and fourteenth postoperative days. If a proximal deep venous thrombosis was detected, anticoagulation was begun immediately. If no proximal thrombosis was observed, the distal thrombosis was considered stable, the serial duplex examinations were terminated, and clinical observation was continued. Of the forty-one patients who had serial examinations, seven (17 per cent) had a proximal deep venous thrombosis in the ipsilateral limb by the fourteenth postoperative day; all seven were asymptomatic, and all were managed with anticoagulation. Thirty-four patients (83 per cent) were determined to have a stable distal deep venous thrombosis on the fourteenth postoperative day, and no additional duplex examinations were performed. Thirty-three (97 per cent) of these thirty-four patients remained asymptomatic for deep venous thrombosis thereafter; the remaining patient (3 per cent) had a proximal deep venous thrombosis in the ipsilateral limb eleven months postoperatively.(ABSTRACT TRUNCATED AT 250 WORDS)