The diagnostic strategy of pulmonary embolism is based firstly on pulmonary scintigraphy, a non-invasive investigation which provides a definitive diagnosis in about 30% of patients, and then on pulmonary angiography, which remains the investigation of reference. However, new diagnostic methods have been introduced in order to reduce the number of angiographies. Measurement of plasma D-dimer, a fibrin degradation product, enables exclusion of the diagnosis in 20-50% of patients without pulmonary embolism when the result is normal on ELISA (< 500 micrograms/l with the commercialized Stago test). This is due to the very high sensitivity of D-dimer: in a compilation of recent series with a total of 1,159 patients suspected of having pulmonary embolism, their concentration was over the threshold of 500 micrograms/l in 96% (CI 95%, 93-98) of patients with pulmonary embolism. On the other hand, their low specificity makes them useless for a positive diagnosis of the condition. Lower limb venous compression ultrasonography enables detection of proximal deep venous thrombosis in about 57% (CI 95%, 52-62) of patients with pulmonary embolism, posing the indication for anticoagulation without further investigations because of its high specificity (98%) (CI 95%, 97-99). When venous ultrasonography is normal, however, pulmonary embolism cannot be excluded. A diagnostic strategy associating these two investigations and pulmonary scintigraphy reduces the number of diagnostic angiographies by 30 to 50% according to whether D-dimer and ultrasonography are performed before or after scintigraphy respectively. More extensive use of D-dimer in clinical practice requires more rapid and equally reliable unitary tests as the ELISA.