The Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study has shown that clinical evaluation and lung scanning may substantiate or exclude pulmonary embolism with reasonable probability in approximately half of the patients in whom it is suspected; for the remainder, pulmonary angiography is considered the gold standard diagnostic test. We performed a decision analysis to assess the potential of two noninvasive tests, D-dimer plasma measurement and lower-limb B-mode venous ultrasound, for reducing the number of pulmonary angiograms necessary to diagnose pulmonary embolism. Our decision model addresses hypothetical patients in the emergency ward with suspected pulmonary embolism and abnormal lung scan results. Results show that D-dimer measurements of less than 500 micrograms/L could be used reliably to exclude pulmonary embolism in patients with an abnormal but not high-probability (inconclusive) lung scan. D-dimer measurements of greater than 500 micrograms/L have no positive predictive value for pulmonary embolism and should be followed by ultrasound, which may replace pulmonary angiography when it discloses deep venous thrombosis. Pulmonary angiography should be performed when ultrasound is negative because of its presumably low sensitivity for deep venous thrombosis in patients with pulmonary embolism. A D-dimer measurement of less than 500 mu/L does not exclude pulmonary embolism in patients with a high clinical suspicion of pulmonary embolism. On the basis of the results of the PIOPED study, we calculated that the combination of D-dimer measurement and ultrasound might reduce the requirement for pulmonary angiography by one third among patients with inconclusive scan results and intermediate clinical probability of pulmonary embolism.