Venography was performed on fifty-six patients suspected of having deep venous thrombosis (DVT) of the legs. The accuracy of the D-dimer measurement in plasma using two latex tests and an enzyme-linked immunosorbent assay (ELISA) was compared with that of usual determination of total fibrin(ogen) degradation products (FDPs) in serum with respect to the presence of DVT. The three D-dimer tests were clearly superior to the FDP assay, but only the ELISA could accurately rule out the diagnosis of DVT with a predictive value of 100% when plasma D-dimer level was less than 200 micrograms/L. However, this test cannot be used for positive diagnosis (false positive rate of 69%). Thus, plasma D-dimer measurement with ELISA allows identification of patients in whom further investigation by means of more specific tests (venography or plethysmography) is indicated in order to establish the diagnosis of DVT. In contrast to this, sensitivity of the two latex tests studied was low (60 and 76%, respectively), which makes them unsuitable for emergency screening. In addition, the potential of D-dimer dosage for diagnosis of DVT in hospitalized patients is hampered by the presence of associated conditions that are responsible for elevated plasma levels in most cases.