BACKGROUND: D-dimer tests are used in various diagnostic strategies to exclude pulmonary embolism (PE). However, their role as an exclusionary first-line test is still uncertain, mainly because accuracy of the test varies according to the assay and the studied population. METHODS: The aim of this multicentre study was to evaluate the accuracy of D-dimer testing in patients with suspected PE. Diagnosis of PE was based on pre-test clinical probability (PCP) evaluation and both single-detector spiral CT (CT) and lower limbs compression ultrasonography (CUS). Lung scanning and/or pulmonary angiography was mandatory when CT or CUS was inconclusive and when both CT and CUS were normal in a patient with a high PCP. All patients were followed-up for 3 months, looking for VTE recurrence. D-dimers were collected within 24 h of inclusion and stored in each local hematology unit, to be analyzed at the end of all inclusions; physicians in charge of the patient were blinded to D-dimer results. RESULTS: Three hundred and fifty two patients were included in 4 centres. Prevalence of PE was 38.6%. PCP was low in 82 (23.3%), intermediate in 176 (50%) and high in 94 (26.7%) patients. Sensitivity of D-dimer was 96.3% (95% CI: 93-99) and negative predictive value reached 94.4% (95% CI: 90-99). Five patients with a confirmed PE had a D-dimer level below 500 ng/ml (two patients with a high PCP). Among 258 patients with low or intermediate PCP, 80 (31%) had a negative D-dimer test result; three of them had a false negative result and the number needed to test was 3.3. Among 94 patients with a high PCP, 9 had a negative D-dimer test result; two of them had a false negative result and the number needed to test was 13.5. CONCLUSION: These results confirm that rapid assays used in this study can safely exclude PE in first-line testing only in non-high CP patients.