A prospective study of the accuracy of real-time B-mode ultrasonography in detecting deep venous thrombosis in the femoral and popliteal veins of the lower extremity was conducted on a consecutive series of patients who had had a total hip replacement. Ascending venography was used as a diagnostic standard. One hundred and forty-three patients had ultrasound studies of both lower extremities and a venographic study of the operatively treated lower extremity at an average of 7.6 days postoperatively. The two tests were done within twenty-four hours of each other. Both ultrasonography and venography were done on 152 extremities. Two paired studies were excluded from the analysis of results because the ultrasound scans could not be interpreted. In 131 extremities, both diagnostic tests were negative for proximal thrombosis. Eight extremities had isolated thrombosis of a vein in the calf that was detected only by venography. In nineteen extremities, old or fresh thrombosis was diagnosed by venography. In four extremities, an old thrombosis of the superficial femoral vein was detected by both studies. In four of the extremities that had a new thrombus in the common femoral vein and in nine that had a new thrombus in the superficial femoral vein, abnormal findings on ultrasound scans correlated with those on venograms. There were two false-negative ultrasound scans and one false-negative venogram. The sensitivity, specificity, and accuracy of ultrasonography were 89, 100, and 99 per cent for the diagnosis of thrombosis of the proximal veins of the lower limb and 63, 100, and 93 per cent for the diagnosis of thrombosis when the entire venous system of the limb was included.(ABSTRACT TRUNCATED AT 250 WORDS)