Twenty-four consecutively hospitalized patients with clinical evidence of lower extremity deep vein thrombosis (DVT) had blinded ipsilateral ultrasonographic duplex imaging (US), contrast venography (VG), and a one-time bilateral impedance plethysmography (IPG) examination. The cases of eight (33%) of the study patients were thought to be complicated due to a prior history of DVT or the presence of ipsilateral inguinal adenopathy. Interpretable test results were obtained for 100% of the US examinations, 88% of the VG studies, and 75% of the IPGs. One additional subject, included in our study for statistical analysis, had an unblinded false-negative US examination; in this case DVT was shown by VG but could not be seen on US despite the radiologist's prior knowledge. The sensitivities, specificities, and 95% confidence intervals for the three studies were as follows: US, 92% (81% to 103%) sensitivity, 100% (99% to 101%) specificity; VG, 100% (99% to 101%) sensitivity, 75% (58% to 96%) specificity; IPG, 50% (30% to 70%) sensitivity, 83% (68% to 98%) specificity. Of the six indeterminate IPG studies, four (67%) occurred in complicated cases. These data suggest that US should be the first-line noninvasive screening study done for evaluation of suspected DVT, especially in complicated cases.