To establish criteria allowing early detection of pathologically significant alterations in pulmonary emphysema caused by smoking, pulmonary-function tests and high-resolution computed tomography were done in 104 subjects categorized into three groups: nonsmoking healthy adults, smokers with a normal FEV1%, and smokers with a low FEV1% (cross-sectional analysis). Fifty-six of the 104 patients underwent pulmonary-function testing and high-resdution computed fomography once per year for 3 years (longitudinal analysis). Cross-sectional and longitudinal analyses showed that abnormalities in functional residual capacity, in single-breath diffusing capacity for carbon monoxide, and in the average tomographic density of sections in the lower lung fields obtained after a deep inspiration could be used to predict whether the disease would reach an advanced stage, even if the patients had no significant symptoms at the time of testing. Relative areas of low-attenuation regions, which were alleged to directly reflect the size of emphysematous areas, appear not to be useful for early detection of pathological emphysema.