In a series of 180 patients, clinically suspected of having deep venous thrombosis (DVT), contrast venography was compared with radionuclide phlebography, duplex ultrasonography and strain gauge plethysmography. In most patients lung scintigraphy was also performed to detect pulmonary embolism (PE). Venography was performed on a routine basis. All venograms were read by at least two observers (radiologists or radiologist/resident) and an inter-observer agreement was reached of 96% with a kappa value of 0.935. In six patients venography was technically impossible or inadequate, 58% of the patients actually had DVT and 26% developed pulmonary embolism (PE). Of the patients with proven DVT, 43% developed PE. Of the three other methods duplex scanning scored the best for the detection of proximal thrombosis, with 92%, 90% and 9.2, and strain gauge plethysmography the worst, with values of 72%, 78% and 3.2 for, respectively, sensitivity, specificity and positive likelihood ratio's. On the basis of the presented material and the current literature it is concluded that the choice for a screening test for proximal thrombosis could best be made on the basis of (local) availability, cost-effectiveness and patient comfort. Duplex ultra-sonography is tipped as the most promising method, accepting that distal thrombosis (calf veins) does not play an important role in PE. Contrast venography should be used as a 'golden backup' in any case of doubt.