This paper reviews the published experience with plasma measurement of D-dimer (DD), a specific degradation product of crosslinked fibrin, in the diagnostic approach of venous thromboembolism (VTE). Pooling 11 studies (with weighting of the figures according to sample size) with a total of 1337 patients clinically suspected of deep venous thrombosis (DVT) (prevalence of DVT 35%) disclosed an average weighted sensitivity of 96.8% (95% CI: 95.2-98.4) and specificity of 35.2% (95% CI: 32.0-38.4) for the presence of DVT when the ELISA technique was used. In 908 patients suspected of pulmonary embolism (PE) from 9 trials (prevalence of PE 38%), the ELISA technique was associated with a weighted sensitivity of 96.8% (95% CI: 95.0-98.6) and specificity of 45.1% (95% CI: 40.8-49.4) for the disease. Figures obtained with latex assays were definitely lower, precluding their use in the diagnostic approach of VTE. These results show that a low concentration of plasma DD measured by the ELISA technique (usually less than 500 micrograms/l) might be used to rule out VTE in clinically suspected patients. Increased plasma concentrations are of no utility because of the low specificity of this test result. The clinical usefulness of the DD ELISA test should now be assessed in management trials under routine conditions, in the frame of clinical decision-making diagnostic processes. Lastly, the promising data obtained in a small number of asymptomatic, postoperative patients at risk of VTE deserve confirmation before the test can be recommended for initial screening in thrombo-prophylactic trials.