Our objective was to investigate possible factors implicated in either early death from or scintigraphic resolution of pulmonary embolism. To that end we conducted a retrospective study of 116 patients with either a high likelihood of pulmonary thromboembolism (PTE) diagnosed by scintiscan or with a fair probability of PTE by scintiscan accompanied by a positive phlebograph. The images were taken upon admission, at 7 days, 10 days and 6 months. The factors analyzed were age, sex, trauma, immobility, surgery, obesity, hemiplegia, venous insufficiency, cardiopulmonary disease, neoplasia, chest X-ray and ECG alterations, D(A-a)O2 and size of perfusion defects upon admission and 7 to 10 days later. We performed single-variable analyses and multiple logical regression analyses using perfusion defect at 6 months as the dependent variable. The early mortality rate (13%) was higher in patients with neoplasms, a larger alveolar-arterial index and greater perfusion defects upon admission. Scintiscans became normal in 28%. Multivariate analysis to predict total or partial resolution at 6 months showed that size of perfusion defects at 7 to 10 days was the best predictive factor. A cutoff point was calculated by analyzing the ROC for this factor. Thus, when the defect at 7 to 10 days was equal to or greater than 1 segment, the probability of residual defects remaining after 6 months was twice as great (sensitivity 83%, specificity 57%). In conclusion, early death was more likely in PTE patients with neoplasms, larger defects upon admission and greater alveolar-arterial difference. Scintigrams showed resolution 6 months after admission in 28%. The size of perfusion defects 7 to 10 days after admission was the factor that best predicted total of partial resolution at 6 months.