The purpose of this investigation was to assess the relative risks of antithrombotic treatment following pulmonary angiography vs no angiography in patients with acute pulmonary embolism (PE). Comparisons of risks of diagnosis and therapy were made among patients treated on the basis of a pulmonary angiographic diagnosis vs patients treated on the basis of a ventilation/perfusion (V/Q) scan combined with clinical assessment in the presence of a single negative test for deep venous thrombosis. Calculations of risks assumed major complications of pulmonary angiography equal to 1.3 percent, major bleeding with heparin equal to 4.9 percent (1.1 percent among patients with a low risk of bleeding and 10.8 percent among patients with a high risk of bleeding), and major bleeding with warfarin (international normalized ratio 2 to 3) equal to 1.7 percent. Among patients with a risk of major bleeding from heparin followed by warfarin of 6.6 percent, if the estimated probability of PE was greater than about 80 percent, fewer major complications of diagnosis and treatment would occur if treatment was initiated on the basis of the V/Q scan. If the probability of PE in such patients was less than 80 percent, fewer major complications of diagnosis and treatment would occur if the diagnosis was established by pulmonary angiography. Among patients with a high (12.5 percent) risk of major bleeding, it was shown to be safer to treat on the basis of an angiographic diagnosis if the estimated probability of PE was less than 90 percent. If the patients, however, were at low (2.8 percent) risk of major bleeding, fewer major complications would occur if angiography was reserved for patients with an estimated risk of PE less than about 50 percent. Serial studies of the leg veins may eliminate the need for angiography in such patients.