Although it is accepted that thoracic structures are enlarged in supine chest radiographs, it is not known whether the lying posture may also influence the radiographic detection of abnormalities, such as those suggestive for pulmonary embolism (PE). For this purpose, we performed planar chest radiographs both in the seated and supine positions, keeping the target-to-film distance at 2 m, in 20 patients with acute PE. Chest radiograph was then repeated 1 month later in either position with the radiological conditions unchanged to investigate the effect of posture on the detection of resolution of signs. The detection of signs due to modifications of preexisting thoracic structures (elevation of the diaphragm and enlargement of the heart, right descending pulmonary artery and azygos vein) in acute PE was different between films taken in the supine and seated positions: both hemidiaphragms were more elevated (p less than 0.001), heart and central vessels were wider (p less than 0.05 or less) when patients were supine. On the contrary, the detection of signs of new occurrences ('sausage'-like appearance of the descending pulmonary artery, linear atelectasis, densities compatible with pulmonary infarction, pleural effusion and focal hyperlucency) was not influenced by the patient's position. During recovery, both kinds of abnormalities were reduced concomitantly in seated and supine radiographs. In conclusion, in acute PE, chest radiographs should be obtained in the seated position since the supine posture may overestimate signs due to modifications of preexisting thoracic structures even though it does not influence the detection of signs of new occurrences. On the contrary, films taken in either position are useful to follow the evolution of PE, provided the film is obtained in the same position and with the same technique as in the acute phase.