This is a prospective study of 108 patients in two distinct groups undergoing real-time ultrasonography (US) and ascending conventional venography within the same day. The two patient groups consisted of the following: Those patients evaluated because of suspicion of deep venous thrombosis of lower limbs (69 patients) and those at high risk for venous thrombosis (19 patients with a recent hip fracture, 20 with a suspected pulmonary embolism). In the diagnosis group 48 patients had venographic evidence of thrombosis. The predictive value of abnormal findings from real-time US was 97%, and that of a negative study was 75%. Thus, real-time US may have a role as a diagnostic procedure, to be followed by x-ray venography in patients with negative US results. By contrast, real-time US is far less sensitive as a screening test in patients without clinical evidence of thrombosis. Only 3 of 9 patients with thrombosis were detected, with a 50% sensitivity for proximal vein thrombosis. Therefore, the use of real-time US for screening high-risk patients must be limited to very high risk patients in whom other tests are ineffective (as in hip surgery).