We compared the age dependence of cross-sectional and longitudinal changes in ventilatory function. FEV1, FVC, and data on chronic respiratory symptoms were obtained from 4,395 adults in a longitudinal survey of normal populations in two different areas in the Netherlands. They participated in up to five surveys at 3-yr intervals between 1972 to 1973 and 1984 to 1985. The ventilatory function in the oldest cohorts is substantially lower than might have been expected from the longitudinal change in the youngest cohorts. This holds for males and females, smokers and nonsmokers, subjects with or without symptoms, and for both survey populations. The robustness of the findings is demonstrated by various data-analytic strategies or omitting one or two of the five surveys from the analysis. Selective loss to follow-up cannot explain the discrepancy. It is concluded that the main differences between our longitudinal and cross-sectional findings may be due to a cohort effect. The implication is that in longitudinal studies of populations at risk, reference equations based on cross-sectional surveys may overestimate longitudinal change and hence lead to underestimating effects of exposure. Similarly, in clinical studies, accelerated decline in ventilatory function may be underrated if it is compared with cross-sectional standards. In older people at any one age the ventilatory function seems to improve in successive birth cohorts.