Mandatory, bilateral, centrally adjudicated venography is currently required to assess the incidence of deep vein thrombosis (DVT) in confirmatory trials of new antithrombotic agents. Non-invasive and less cumbersome methods would be welcomed by both patients and investigators. Centrally adjudicated complete compression ultrasound (CCUS) of all leg veins could become an alternative to venography. Despite the fact that no formal validation of venography has ever been performed, CCUS will have to be externally validated against this 'gold standard' to gain general acceptance and regulatory approval. A substudy of two recent phase IIb trials of a novel, oral anticoagulant for the prevention of venous thromboembolism in patients undergoing elective hip or knee arthroplasty was undertaken to validate CCUS against venography. The studies had similar designs and methodology, and were designed to allow the pooling of data. Mandatory bilateral venography was performed 7±2 days after surgery; CCUS was performed later the same day, with the sonographers blinded to the venography result. All sonographers were required to have completed a training course, and to have obtained personal certification for the standardized examination and documentation procedure. Venograms and CCUS videos were adjudicated centrally by two independent readers at separate locations; discrepancies were resolved by consensus reading. From the 1347 patients participating in the two trials, 870 matching pairs of evaluable venograms and CCUS videos were obtained. Observed prevalences of any DVT, including muscle vein thrombosis, were 19.1% with venography, and 13.0% with CCUS. Sensitivity and specificity values [with 95% confidence intervals (95% CI)] for detecting DVT with CCUS are shown in the table. Although false positives did not severely impair specificity, the small number of true positives detected by CCUS resulted in remarkably low sensitivity, in particular for proximal DVT. Based on these external validation figures, centrally adjudicated CCUS is not a viable technique to replace venography for the screening of DVT early after major orthopaedic surgery in confirmatory trials of novel antithrombotic agents.DVTSensitivity [% (95% CI)]Specificity [% (95% CI)]Any27.4 (21.2, 33.6)95.8 (94.8, 96.8)Proximal13.0 (−0.7, 26.8)99.2 (98.9, 99.7)Distal25.9 (19.7, 32.1)96.2 (95.3, 97.2)