The relationship between smoking habits, respiratory symptoms and lung function at the start of the study and mortality during a follow-up period of 11 years was studied in 607 men, aged 50 and 60 years at entry and sampled from the general population. The overall mortality rate in the sample was 18%. In a logistic multiple regression model, mortality rate was significantly related to age, smoking habits, dyspnea and one of the lung function variables FEV1, VC or the slope of phase III. Smokers had a double mortality rate compared to non-smokers (22 versus 10%) after allowing for age, dyspnea and lung function. Similarly, in subjects with abnormal FEV1, VC or slope of phase III, the mortality rate was almost doubled compared to subjects with normal lung function, other factors being equal. Thus, impaired lung function is an important factor to be considered in the assessment of mortality risk, besides smoking and dyspnea.