BackgroundMeasurement of expired-air carbon monoxide (EACO) is commonly used to ascertain non-smoking status, although it can also reflect exposures not related to smoking. Our aim was to assess 16-year mortality according to EACO measured at baseline, in a general population. MethodsOur analysis was based on the Third French MONICA population survey (1994–1997). Causes of death were obtained 16years after inclusion, and assessment of determinants of mortality was based on Cox modeling. ResultsEACO was measured in 2232 apparently healthy participants aged 35-64. During follow-up, 195 deaths occurred (19% were due to cardio-vascular (CV) causes and 49% to cancer). At baseline, the mean EACO was 11.8 (±7.4)ppm, 4.6 (±2.5)ppm, 4.3 (±2.2)ppm for current, former and never smokers, respectively (P<0.001). After adjustment for main mortality risk factors and smoking, the hazard ratio (HR) for total mortality was 1.03[95% confidence interval: 1.01–1.06] per 1-unit increase in EACO, and it was 1.04[1.01–1.07] for cancer mortality. Adjusted HR for CV mortality was 1.05[1.01–1.10] but did not remain significant after additional adjustment for smoking (0.98[0.91–1.04]). Interactions between EACO and smoking were not significant. ConclusionsIn a general population, baseline EACO is an independent predictor of 16-year all-cause and cancer mortality, after adjustment for confounders including smoking. Given that the effect of EACO is similar among smokers and non-smokers, EACO is probably not solely related to smoking but could also be a marker of inhaled ambient carbon monoxide and/or endogenous production. Besides, smoking better predicts CV mortality than EACO.