Background and objective Smoking increases perioperative complications. Carbon monoxide concentrations can estimate patients' smoking status and might be relevant in preoperative risk assessment. In smokers, we compared measurements of carbon monoxide in expired air (COexp) with measurements of carboxyhaemoglobin (COHb) in arterial blood. The objectives were to determine the level of correlation and to determine whether the methods showed agreement and evaluate them as diagnostic tests in discriminating between heavy and light smokers. Methods The study population consisted of 37 patients. The Micro Smokerlyzer was used to measure COexp; it measures COexp in parts per million (ppm) and converts it to the percentage of haemoglobin combined with carbon monoxide (%Hb). COHb in arterial blood was measured by the ABL 725. Correlation analysis and Bland-Altman analysis were performed, and 2 x 2 contingency tables and receiver operating characteristic curve analysis were conducted. Results The correlation between the methods was high (rho = 0.964). Bland-Altman analysis demonstrated that the Micro Smokerlyzer underestimated COHb values. The areas under the receiver operating characteristic curves were 0.746 (ABL 725) and 0.754 (Micro Smokerlyzer) and, by comparison, no statistically significant difference was found (P = 0.815). Conclusion The two methods showed a high level of correlation, but poor agreement. The Micro Smokerlyzer systematically underestimated COHb values and, in order to avoid this, we suggested an alternative algorithm for converting COexp from ppm to %Hb. The ABL 725 and Micro Smokerlyzer were fair diagnostic tests in distinguishing between heavy and light smokers, but the longer the patients' smoking cessation time, the poorer the ability as diagnostic tests. Eur J Anaesthesiol 2010;27:812-818