One hundred twenty-six patients with clinically suspected acute deep venous thrombosis of the lower extremity (DVT) were examined comparatively with ultrasound and venography. In total, 174 lower extremity venograms were obtained. Ultrasonic examinations were performed on patients in the supine position. The venous segments were evaluated almost exclusively with transversal scanning. In the thigh, the only criterion for DVT was the reduced or absent compressibility of the venous lumen when gently compressed with the transducer. In the calf, normal unobstructed veins can usually not be viewed in the supine patient, whereas thrombotic veins appear as sonolucent, incompressible channels. Eight-three of the 174 lower extremity venograms were positive for DVT. In the majority of cases (53 of 83) the thrombotic process had involved two or more segments in combination. The sites of involvement of the different venous segments were distributed as follows: 24 occlusions of the common femoral vein, 52 of the superficial femoral vein, 56 of the popliteal vein, and 71 of the calf veins. Ultrasound had a sensitivity of 100% for thrombosis of the common femoral vein, 96% for the superficial femoral veins, 98% for the popliteal vein, and 93% for the calf veins. For the entire lower extremity, in regard to the diagnosis of thrombosis, the overall sensitivity was 95%. In 90% the extension of the occlusion was foreseen correctly. In no cases were false-positive results reported. Thus the overall specificity was 100%. The authors conclude that real-time ultrasound is a highly accurate method for the diagnosis of DVT of the lower extremity. It is the only indirect method capable of evaluating the venous system of the thigh, as well as that of the calf, with high accuracy. It should be the first choice of diagnostic imaging method in the diagnosis of deep venous thrombosis of the lower extremity.