OBJECTIVE: To evaluate the usefulness of B-type natriuretic peptide and troponin I measurements in predicting right ventricular dysfunction (RVD) in non-massive pulmonary embolism. DESIGN: Prospective observational study. SETTING: University-affiliated emergency unit, cardiology and pneumology departments. PATIENTS: Sixty-seven patients admitted because of acute pulmonary embolism, without shock on admission, completed the study. INTERVENTIONS: Blood samples and echocardiography were obtained on admission for subsequent and independent assessment of B-type natriuretic peptide (BNP) and troponin I levels as well as RVD. MEASUREMENTS AND RESULTS: Echocardiographic RVD was diagnosed in 36 patients and was severe in 13 on admission. BNP and troponin I levels were higher in patients with RVD than in those with no RVD [62 (27-105) vs. 431 (289-556) pg/ml for BNP, p<0.001; 0.01 (0-0.09) vs. 0.16 (0.03-0.32) microg/l for troponin I, p=0.005]. The area under the receiving operating characteristic curve (AUC) for diagnosing RVD was 0.93 for BNP and 0.72 for troponin I. The troponin I level increased further when RVD was severe, compared with moderate, and the AUC was 0.91 for identifying severe RVD. Diagnoses of RVD and severe RVD were ruled out by BNP<or=100 pg/ml (30% of patients) and troponin I 0.10 microg/l (58% of patients), respectively. In-hospital death or circulatory failure occurred in nine patients; all had echographic RVD and level of BNP >100 pg/ml and troponin I >0.10 microg/l. CONCLUSION: In hemodynamically stable pulmonary embolism, BNP/troponin I measurement is helpful on admission, especially for ruling out RVD, i.e. patients with in-hospital high-risk.