The aim of this study was to prospectively evaluate the accuracy of real-time B-mode ultrasonography in the diagnosis of asymptomatic proximal deep vein thrombosis (DVT) in hip surgery patients. Venography was adopted as the gold standard. We studied 100 consecutive patients undergoing hip surgery: 60 patients for hip fracture and 40 patients for elective hip replacement. Bilateral real-time B-mode ultrasonography was performed prior to bilateral venography. The two diagnostic procedures were performed on the same day by different investigators unaware of the assigned prophylatic regimen for DVT. Compressibility of the vein segment was adopted as the single criterion for DVT. Venography was performed and judged by radiologists unaware of the ultrasonography results. In 13 limbs venography was either impossible to perform or not adequate for judgement. Ultrasonography and an adequate venography was obtained in 187 limbs. A venography proven DVT was observed in 49 limbs (26.2%) and a proximal DVT in 21 limbs (11.2%). All the patients were asymptomatic for DVT. The sensitivity and specificity of real time B-mode ultrasonography for proximal DVT were 57% (95% confidence interval: C.I. 36-80) and 99% (C.I. 99-100), respectively and the positive and negative predictive values were 93% (C.I. 73-100) and 95% (C.I. 91-97), respectively. The sensitivity and specificity for overall DVT were 25% (C.I. 11-38) and 99% (C.I. 97-100), respectively and the positive and negative predictive values were 92% (C.I. 73-100) and 79% (C.I. 76-85), respectively. Our data indicate that real-time B-mode ultrasonography for its high specificity could make venography unnecessary in patients with positive results.(ABSTRACT TRUNCATED AT 250 WORDS)