BACKGROUND AND OBJECTIVE: Pulmonary embolism (PE), with an incidence of 23 per 100,000 patients per year, is a frequent clinical problem, responsible for 200,000 deaths each year in the United States. Pulmonary angiography, the gold standard for diagnosing PE, is invasive, costly and not universally available. Moreover, PE is confirmed in only approximately 30% of patients in whom it is suspected, rendering noninvasive screening tests necessary. Several strategies have been recently proposed to reduce the need for pulmonary angiography in the diagnostic workup of pulmonary embolism. The objective of this article is to analyze the individual performance of the new diagnostic instruments and their combination in rational diagnostic strategies. METHODS: The author has been working in this field and has contributed original papers on diagnosis of pulmonary embolism and cost-effectiveness of noninvasive diagnostic tests. In addition, the material examined in this article includes articles published in the journals covered by the Science Citation Index and Medline. RESULTS: Several strategies have been recently proposed to reduce the need for pulmonary angiography in the diagnostic workup of pulmonary embolism. The PIOPED study has established the value of ventilation-perfusion lung scan, a normal perfusion lung scan virtually ruling out PE, whereas a high probability lung scan is considered diagnostic in face of reasonable clinical suspicion. All other lung scan results are nondiagnostic. However, clinical evaluation, although insufficiently accurate to yield a definitive diagnosis, is probably reliable enough to be used for estimating pretest probability of PE. The combination of a low clinical probability of PE and a so-called low probability lung scan yields a very low posttest probability of PE, thus foregoing the need for pulmonary angiography. Other useful instrument in patients with nondiagnostic scans is plasma D-dimer (DD) measurement (ELISA assay), which when under a cutoff value of 500 micrograms/L potentially exclude PE, due to high sensitivity (97%). Conversely, venous compression ultrasonography of the lower limbs (US) is highly specific (98%) for deep vein thrombosis (DVT), and disclosing a DVT warrants anticoagulant treatment without resorting to angiography. The potential role of echocardiography is also discussed. The rational sequence of noninvasive tests is currently under discussion. Performing D-dimer and US before lung scan may be the most cost-effective strategy, pulmonary angiography being performed only in case of an inconclusive noninvasive workup. INTERPRETATION AND CONCLUSIONS: Even though PE remains a difficult diagnostic challenge, the availability of novel noninvasive tests (plasma D-dimer and ultrasonography of the lower limbs) and the rehabilitation of clinical assessment allow a more rational and sparse prescription of pulmonary angiography. More work needs to be done to assess test performances and refine diagnostic strategies in distinct patient subgroups, particularly those hospitalized. Screening patients with plasma D-dimer and ultrasonography of the lower limbs may be the most cost-effective strategy, at least in outpatients.