Background:  Plasma B‐type natriuretic peptide (BNP) assay is recommended as a diagnostic tool in emergency‐room patients with acute dyspnea. In the intensive care unit (ICU), the utility of this peptide remains a matter of debate. The objectives of this study were to determine whether cut‐off values for BNP and N‐terminal‐proBNP (NT‐proBNP) reliably diagnosed right and/or left ventricular failure in patients with shock or acute respiratory distress, and whether non‐cardiac factors led to an increase in these markers. Methods:  Plasma BNP and NT‐proBNP levels and echocardiographic parameters of cardiac dysfunction were determined in 41 patients within 24 h of the onset of shock or acute respiratory distress. Results:  BNP and NT‐proBNP levels were higher in the 25 patients with heart failure than in the other 16 patients: 491.7 ± 418 pg/ml vs. 144.3 ± 128 pg/ml and 2874.4 ± 2929 pg/ml vs. 762.7 ± 1128 pg/ml, respectively (P < 0.05). In the diagnosis of cardiac dysfunction, BNP > 221 pg/ml and NT‐proBNP > 443 pg/ml had 68% and 84% sensitivity, respectively, and 88% and 75% specificity, respectively, but there was a substantial overlap of BNP and NT‐proBNP values between patients with and without heart failure. BNP and NT‐proBNP were elevated, but not significantly, in patients with isolated right ventricular dysfunction. Patients with renal dysfunction and normal heart function had significantly higher levels of BNP (258.6 ± 144 pg/ml vs. 92.4 ± 84 pg/ml) and NT‐proBNP (2049 ± 1320 pg/ml vs. 118 ± 104 pg/ml) than patients without renal dysfunction. Conclusion:  Both BNP and NT‐proBNP can help in the diagnosis of cardiac dysfunction in ICU patients, but cannot replace echocardiography. An elevated BNP or NT‐proBNP level merely indicates the presence of a ‘cardiorenal distress’ and should prompt further investigation.