Deep venous thromboembolism and pulmonary embolism are still underdiagnosed in the ICU. Thromboelastography (TEG) has shown considerable variability in sensitivity and specificity as a predictor of venous thromboembolism (VTE). We designed a prospective double-blind observational study to predict the risk of VTE using TEG in a cohort of critically ill patients. Seventy-two hours after admission in the ICU and consequent prophylaxis with low-molecular-weight heparin, we performed compressive color-Doppler ultrasound and diagnosed deep venous thrombosis. Computed tomography scan was performed for the diagnosis of pulmonary embolism if pulmonary embolism was suspected based on physical examination and transthoracic echocardiography. Whole blood samples were obtained from central venous lines 6-8 h after subcutaneous administration of low-molecular-weight heparin. Native TEG and modified heparinase TEG were performed using a Thromboelastograph Coagulation Analyzer. Fifty-seven patients were consecutively enrolled of which six (10.5%) developed deep venous thrombosis; two (3.5%) also developed pulmonary embolism. The native thrombodynamic ratio (TDR) was an independent predictor of the odds of thrombosis (odds ratio 1.016, P < 0.05, 95% confidence interval 1.008-1.047), with a 0.93 area under the ROC curve. Using 10.6 as the lower cut-off point, TDR showed 100% sensitivity and 0 negative likelihood ratio (95% confidence interval 0-0.4) in excluding the clinical diagnosis of VTE. Our results show that TDR predicts VTE in the ICU. Our findings are in agreement with those reported by other investigators, who demonstrated that a TDR less that 10 is associated with prophylactic levels of anti-Xa.