Objective: To identify potentially confounding variables for the interpretation of plasma N-terminal pro brain natriuretic peptide (NT-proBNP).

Design: Randomly selected subjects filled in a heart failure questionnaire and underwent pulse and blood pressure measurements, ECG, echocardiography, and blood sampling.

Setting: Subjects were recruited from four Copenhagen general practices located in the same urban area and were examined in a Copenhagen University Hospital.

Patients: 382 women and 290 men in four age groups: 50–59 years (n = 174); 60–69 years (n = 204); 70–79 years (n = 174); and > 80 years (n = 120).

Main outcome measures: Associations between the plasma concentration of NT-proBNP and a range of clinical variables.

Results: In the undivided study sample, female sex (p < 0.0001), greater age (p < 0.0001), increasing dyspnoea (p = 0.0001), diabetes mellitus (p = 0.01), valvar heart disease (p = 0.002), low heart rate (p < 0.0001), left ventricular ejection fraction ⩽ 45% (p < 0.0001), abnormal ECG (p < 0.0001), high log₁₀[plasma creatinine] (p = 0.0009), low log₁₀[plasma glycosylated haemoglobin A1c] (p = 0.0004), and high log₁₀[urine albumin] (p < 0.0001) were independently associated with a high plasma log₁₀[plasma NT-proBNP] by multiple linear regression analysis.

Conclusions: A single reference interval for the normal value of NT-proBNP is unlikely to suffice. There are several confounders for the interpretation of a given NT-proBNP concentration and at the very least adjustment should be made for the independent effects of age and sex.