Introduction. Brain natriuretic peptide (BNP) and N-Terminal pro natriuretic peptide (NT-proBNP) are widely accepted to diagnose congestive heart failure (CHF) in the emergency room. The aim of this study was to evaluate the value of BNP and NT-proBNP to diagnose CHF in primary care. Methods. Clinical and Doppler-echocardiographic assessment of patients referred by their general practitioner (GP) with the diagnosis of CHF. Receiver operating curves were used to evaluate the accuracy of BNP and NT-proBNP for echocardiographically confirmed systolic and/or diastolic heart failure. Results. Three hundred and eighty four patients (mean age of 65) were included. One hundred and ninety three (50%) patients had systolic heart failure and 31 (8%) had isolated diastolic heart failure. Using currently recommended cut-off values of BNP (less than 100pg/ml) and NT-proBNP (less than 125pg/ml) for exclusion of CHF, BNP was false negative in 25% and NT-proBNP in 10% of the patients. The area under the curve was better for NT-proBNP than for BNP (0.742 vs. 0.691). Conclusion. In this population with a high prevalence of CHF, BNP and NT-proBNP failed to adequately rule out CHF. GP's should be cautious when using BNP and NT-proBNP in primary care. An echocardiography remains compulsory in unexplained dyspnea.