BACKGROUND AND OBJECTIVES: The measurement of D-dimer is claimed to have potential value in excluding deep vein thrombosis (DVT). New rapid methods have been proposed, but few clinical trials have assessed their performance in an emergency context. The different accuracies found between the D-dimer assays have been related to the test used (latex or ELISA), but other variables (such as population investigated, thrombus extension, duration of symptoms or concomitant heparin treatment) may be important, even if not sufficiently investigated. DESIGN AND METHODS: We evaluated the accuracy of a rapid semi-quantitative D-dimer test (Dimertest, Dade Behring), with reference to: a) its use at an emergency unit; b) concomitant heparin administration; c) location of venous thrombosis (VT) (in the deep or superficial venous system limited to the great saphenous vein) and d) symptoms older than 14 days. RESULTS: Two hundred and ninety-eight patients suspected of having DVT and 116 suspected of thrombosis of the great saphenous vein (GSV) were investigated. In the DVT patients, the sensitivity, specificity, positive and negative predictive values were 77.4% (95% CI 68.9-85.9), 81.4% (95% CI 76.1-86.7), 65.4% (95% CI 56.5-74.3) and 88.8% (95% CI 84.2-93.4), respectively. Excluding patients receiving heparin and those with symptoms older than 15 days, the sensitivity and negative predictive value increased to 86.3% (95% CI 78.4-94.2) and 92.8% (95% CI 88.4-97.2), respectively. In patients with GSV thrombosis, the sensitivity, specificity, positive and negative predictive values were 48% (95% CI 34.5-61.5), 90.6% (95% CI 83.2-97.9), 80.6% (95% CI 66.6-94.6) and 68.2% (95% CI 57.8-78.6), respectively. Excluding patients receiving heparin and those with symptoms older than 15 days, did not change the sensitivity or negative predictive value significantly. INTERPRETATION AND CONCLUSIONS: Our results show that previous or concomitant heparin administration, non-acute symptoms and thrombosis localized to superficial veins reduce the clinical usefulness of the D-dimer test as the rate of false negative results is increased.