We assessed the accuracy of venous compression ultrasonography (CUS) for the detection of asymptomatic deep vein thrombosis in 122 consecutive patients, with a mean age of 69 years, who were hospitalised in an internal medicine unit. All included patients had CUS within 48 h of admission. Twelve out of 17 patients with a positive CUS underwent phlebography, as the others withdrew their consent, whereas the remaining 105 patients with a negative serial CUS testing were clinically followed-up at 3 months. We found that CUS had a sensitivity and a specificity of respectively 1 (95% CI, 0.73 to 1) and 1 (95% CI, 0.96 to 1) for the detection of asymptomatic deep vein thrombosis. Kappa-coefficients for intra-observer and inter-observer agreements were respectively 0.88 and 0.56. We concluded that venous compression ultrasonography, performed as described, fulfils requirements of a screening test that could be available for prophylactic clinical trials or epidemiological researches.