Background B-type natriuretic peptide (BNP) levels predict cardiovascular risk in several settings. We hypothesized that they would identify individuals at increased risk of early cardiac complications after major non-cardiac surgery. The current study tests this hypothesis. Methods Two hundred and four patients undergoing major non-cardiac surgery were studied. The primary end-point was the development of acute myocardial injury [defined as cardiac troponin I (cTnI) level > 0.32 ng ml−1] or death in the 3 days after surgery. Results Preoperative BNP levels were raised in patients who died or suffered perioperative myocardial injury (median 52.2 vs 22.2 pg ml−1, P = 0.01) and BNP predicted this outcome with an area under the receiver operating characteristic curve of 0.72 [95% confidence interval (CI) 0.59–0.86, P = 0.01]. A preoperative BNP value > 40 pg ml−1 was associated with an increased risk of death or perioperative myocardial injury [odds ratio (OR) 6.8, 95% CI 1.8–25.9, P = 0.003], and remained independently predictive after correction for the Revised Cardiac Risk Index. Preoperative BNP levels were higher in patients who exhibited new onset atrial fibrillation or ST/T-wave changes on their postoperative ECG (median 50.5 vs 22.5 pg litre−1, P = 0.01). They were also higher in patients who had either elevation of cTnI > 0.32 ng ml−1 or postoperative ECG abnormalities (median 50.4 vs 21.5 pg ml−1, P Conclusions In the setting of major non-cardiac surgery, preoperative BNP levels are higher in patients who experience perioperative death and myocardial injury. Larger studies are required to confirm these data and to clarify what BNP levels may add to existing methods of risk stratification.