OBJECTIVE: We compared the usefulness of plasma N-terminal-pro B-type natriuretic peptide and troponin I levels for risk stratification of patients with pulmonary embolism. METHODS: This was a prospective study performed in an emergency department. N-terminal-B-type natriuretic peptide assay and troponin I were performed blindly at admission in patients with pulmonary embolism confirmed by imaging tests. A complicated pulmonary embolism was defined as any of the following: death, cardiopulmonary resuscitation, requirement for mechanical ventilation, use of pressors, thrombolysis, surgical embolectomy or admission in an intensive care unit. RESULTS: Sixty patients (mean age+/-standard deviation of 72+/-15 years) were included. Seventeen (28%) patients had adverse events: all were admitted in intensive care unit, one was treated with surgical embolectomy and one with thrombolysis, and three died. The median N-terminal-pro B-type natriuretic peptide level (95% confidence interval) was higher in the group of patients with complicated pulmonary embolism, 4086 pg/ml (505-8998) versus 352 pg/ml (179-662), respectively (P<0.05). The mean value of troponin I was similar in the complicated pulmonary embolism group, 0.09+/-0.17 microg/l versus 0.08+/-0.41 microg/l, respectively (P=0.93). The best threshold value of N-terminal-pro B-type natriuretic peptide was 1000 pg/ml, and the receiver operating characteristic curve demonstrated that N-terminal-pro B-type natriuretic peptide significantly predicted the complicated pulmonary embolism with an area under the receiver operative curve of 0.72 (0.58-0.83) (P<0.05), whereas troponin I did not [area under the receiver operative curve of 0.58 (0.42-0.71)]. CONCLUSION: Unlike troponin I, N-terminal-pro B-type natriuretic peptide may be an accurate marker of in-hospital complication after pulmonary embolism.