Reliable prediction of adverse outcomes in acute pulmonary embolism may help choose between in-hospital and ambulatory treatment. We aimed to identify predictors of adverse events in patients with pulmonary embolism and to generate a simple risk score for use in clinical settings. We prospectively followed 296 consecutive patients with pulmonary embolism admitted through the emergency ward. Logistic regression was used to predict death, recurrent thromboembolic event, or major bleeding at 3 months. Thirty patients (10.1%) had one or more adverse events during the 3-month follow-up period: 25 patients (8.4%) died, thromboembolic events recurred in 10 patients (3.4%), and major bleeding occurred in 5 patients (1.7%). Factors associated with an adverse outcome in multivariate analysis were cancer, heart failure, previous deep vein thrombosis, systolic blood pressure <100 mmHg, arterial PaO2 <8 kPa, and presence of deep vein thrombosis on ultrasound. A risk score was calculated by adding 2 points for cancer and hypotension, and 1 point each for the other predictors. A score of 2 best identified patients at risk of an adverse outcome in a receiver operating characteristic curve analysis. Of 180 low-risk patients (67.2%) (score < or =2), only 4 experienced an adverse outcome (2.2%), compared to 23 (26.1%) of 88 high-risk patients (score > or =3). A simple risk score based on easily available variables can accurately identify patients with pulmonary embolism at low risk of an adverse outcome. Such a score may be useful for selecting patients with pulmonary embolism eligible for outpatient care.