Vital capacity, FEV1 and various measurements of forced expiratory flows as well as closing volume (CV), closing capacity (CC), the mean slope of the alveolar plateau (phase III), the maximal amplitude of the cardiogenic oscillations (N2-osc), and other variables obtained from the nitrogen CV-test were measured in random samples of 50- (n = 221) and 60-year-old (n - 410) men in Gothenburg. All the variables were significantly different in smokers consuming 15 g or more tobacco daily compared with asymptomatic non-smokers. Reference values were established from measurements in the non-smokers without respiratory symptoms, and the sensitivities of the measured variables to tobacco smoking were assessed. Phase III revealed the highest sensitivity, being abnormal in about 40-60 % of subjects smoking 15 g or more tobacco daily and in about 30-35 % of subjects smoking less. CV and FEV1 were approximately equally sensitive, being abnormal in about 10-30 % of subjects smoking 15 g or more daily and in about 5-20 % in subjects smoking less. Other measurements of maximal expiratory flow rates gave values below minus two standard deviations for the appropriate reference group in less than 15 % of the smokers. However, when we took the skewed distribution of reference values into account we found a two- to sixfold increase of the sensitivities of these variables in the 50-year-old smokers, in which abnormal values were obtained in about 30-40 % of subjects smoking 15 g or more daily and in about 10-15 % in subjects smoking less. Taking all variables obtained from the CV-test in conjunction, one or more abnormal valves were obtained in 60-70 % of the subjects smoking 15 g or more daily. To the extent that sensitivity to smoking reflects sensitivity to abnormalities in small airways, phase III appeared to be the most efficient single measurement for detection of this condition in the present age groups.