Objectives: To establish the accuracy and identify advantages and disadvantages of colour duplex imaging compared with venography in the diag nosis of symptomatic proximal deep vein throm bosis (DVT) in our centre. Design: Prospective study over a 15 month period. Patients: 34 patients with 35 DVT suspected lower limbs were studied. They were first studied with colour duplex ultrasonography and the findings compared with contrast venography of the suspect limb. Results: 17 venograms were positive for DVT and ultrasound was positive in all. There was one false positive ultrasound examination. Sensitivity and specificity for detection of proximal DVT was 100% and 94.4% respectively. Venography resulted in complications in 6 patients (phlebitis 4, skin ulcer ation 2). Additional diagnoses were possible with sonography (calf haematoma 1, calf neoplasm 1, deep venous reflux 2 and superficial venous reflux 5). Conclusion: Colour coded duplex sonography is highly accurate and safe in detecting femoropopliteal thrombosis. Venography should be reserved only for duplex inconclusive situations.