The potential of plasma measurement of D-dimer (DD), a specific derivative of crosslinked fibrin, for diagnosis or exclusion of pulmonary embolism was investigated in a prospective series of 171 consecutive patients who attended an emergency department with suspected pulmonary embolism. The diagnosis was made or excluded by means of a clinical decision-making process which included clinical evaluation, ventilation-perfusion (VQ) lung scan, and, as indicated, pulmonary angiography, venography, or non-invasive examination of the leg veins. Pulmonary embolism was diagnosed by this process in 55 (32%) of 170 patients with sufficient data. All but 1 of these 55 patients had a DD concentration of 500 micrograms/l or above. The sensitivity and specificity of this cutoff concentration for the presence of pulmonary embolism were 98% and 39%, respectively, which give positive and negative predictive values of 44% and 98%. Among the 115 patients (68%) who had inconclusive VQ scans, 31 were diagnosed as having pulmonary embolism. 29 of the remaining 84 patients without pulmonary embolism had DD concentrations below 500 micrograms/l, which means that further diagnostic procedures could have been avoided in a quarter of the patients with inconclusive VQ scans. The sensitivity of the plasma measurement of DD remained high even 3 and 7 days after presentation (96% and 93%). Plasma measurement of DD therefore has a definite place in the diagnostic procedure for suspected acute pulmonary embolism in attenders at emergency departments: a concentration below 500 micrograms/l rules out the diagnosis.