D-dimer fragments can be measured easily in plasma and whole blood, and the presence or absence of D-dimer could be useful in the diagnostic evaluation of venous thromboembolism. We systematically reviewed the English literature for articles that compared D-dimer results with those of other tests for deep venous thrombosis or pulmonary embolism. Twenty-nine studies were selected for detailed review, and we noted wide variability in assay performance, heterogeneity among subjects, and failure to define absence or presence of venous thromboembolism by a comprehensive criterion standard for diagnosis. These methodologic problems limit the generalizability of the published estimates of D-dimer accuracy for deep venous thrombosis or pulmonary embolism, and the clinical utility of this potentially important test remains unproved.