D-dimers are specific fibrin degradation products which have been measured in plasma and used as a diagnostic test for deep vein thrombosis of the lower limbs. However, the results published so far have often been conflicting. We compared the results of D-dimer assays by the latex method (positivity threshold: 0.5 microgram/ml) in 3 groups of patients: 22 patients with clinically suspected thrombophlebitis confirmed by doppler ultrasound; 22 patients with clinically suspected thrombophlebitis not confirmed by doppler ultrasound; and 17 hospital patients with other diseases serving as controls. The sensitivity of a D-dimer concentration higher than 0.5 microgram/ml was 96 percent and its specificity 69 percent. In an internal medicine unit where the prevalence of thrombophlebitis was 50 percent, the positive predictive value of a D-dimer concentration higher than 0.5 microgram/ml was 81 percent, and its negative predictive value was 94 percent. Thus, because of its low specificity as compared with imaging methods the D-dimer assay test cannot be used to diagnose deep vein thrombosis, but as it has good sensitivity and is rapidly performed, safe and inexpensive it can be used as an emergency examination which, when negative, makes this diagnosis unlikely.