SummaryTo determine their ability to diagnose postoperative deep vein thrombosis (DVT) D-dimer – by three methods – fibrinogen degradation products (FgDP) and fibrinogen levels were measured in 68 consecutive patients before elective surgery for hip replacement and on postoperative day 1, 3, 6, and 10. All patients received prophylaxis and underwent compression real-time B-mode ultrasonography (C-US) on postoperative day 5 and 9, and bilateral ascending venography on day 10. Twenty-two out of 68 patients developed asymptomatic postoperative DVT, which was limited to the calf veins in 14 and involved the proximal veins in 8 patients. C-US was negative in all patients on day 5. On day 9, C-US sensitivity and specificity for proximal DVT were 63% (95% confidence interval: 26%-90%) and 98% (89%-100%) respectively. Postoperative changes in the laboratory parameters evaluated were not different in patients with or without DVT until day 10. On day 10, mean D-dimer, FgDP and fibrinogen levels were significantly higher in patients with DVT than in those without DVT (p values between 0.006 and 0.032), but only D-dimer was higher with DVT involving two or more venous segments than with thrombosis involving one venous segment only (p <0.05). Stepwise logistic regression analysis identified D-dimer and fibrinogen on day 10 as predictors of postoperative DVT. In a receiver operator curve and after weighing for the coefficients generated by logistic regression analysis, the combination of a latex photometric immunoassay and of PT-derived fibrinogen yielded - at a cut-off value of 7.0 - a sensitivity of 100% (73%-100%) and a specificity of 58% (39%-75%) for DVT, with a negative predictive value of 100% (78%-100%), a positive predictive value of 52% (32%-71%) and an overall accuracy of 71% (55%-83%). These results suggest that two simple, fast and reproducible tests may permit the identification of patients at low risk of having postoperative DVT and that a combination of sensitive laboratory assays and of the highly specific C-US may select patients requiring anticoagulant treatment. Efficacy and cost-effectiveness of this approach should be evaluated in large clinical management studies.