Compared with conventional duplex imaging, color-flow scanning facilitates the identification of veins (especially below the knee), decreases the need to assess Doppler flow patterns and venous compressibility, and allows veins to be surveyed longitudinally. These advantages translate into a less demanding and time-consuming examination. This study was designed to determine the accuracy of color-flow scanning for detecting acute deep venous thrombosis in patients in whom the diagnosis is clinically suspected and in asymptomatic patients at high risk for developing postoperative deep venous thrombosis. The diagnostic group included 77 limbs of 75 patients, and the surveillance group included 190 limbs of 99 patients undergoing total hip or knee replacement. All patients were prospectively examined with color-flow scanning and phlebography. In the diagnostic group, the incidence of thrombi in below-knee veins (47%) was approximately equal to that in above-knee veins (43%); but in the surveillance group, the incidence of thrombi in below-knee veins (41%) far exceeded that in veins above the-knee (3%). Nonocclusive clots and clots isolated to a single venous segment were more common in the surveillance group. In symptomatic patients, color-flow scanning was 100% sensitive and 98% specific above the knee and 94% sensitive and 75% specific below the knee. In the surveillance group, color-flow scanning was significantly (p less than 0.001) less sensitive (55%) for detecting thrombi, 93% of which were confined to the tibioperoneal veins. Negative predictive values were 100% and 88% for the diagnostic and surveillance limbs, respectively. Positive predictive values were 80% for the diagnostic limbs and 89% for the surveillance limbs. Color-flow scanning effectively excludes above-knee deep venous thrombosis in symptomatic patients and asymptomatic high-risk patients and predicts the presence of above-knee thrombi in patients in the diagnostic group with reasonable accuracy (97%). We conclude that color-flow scanning is as accurate as conventional duplex imaging and, because of its advantages, is the noninvasive method of choice for evaluating patients with suspected deep venous thrombosis. Its role in the surveillance of patients at high risk remains to be determined and awaits further clinical evaluation.