Longitudinal changes in FEV1 (delta FEV1) have been examined in a sample of the general population of Tucson, Arizona. Adults with satisfactory spirometric tests in all 7 surveys performed over a period of 11 yr showed some between-survey differences in mean FEV1. Although small in absolute magnitude, such survey biases (which would be impossible to detect unless multiple spirometric tests were carried out on a relatively large group of subjects) could have a marked effect on calculated changes in FEV1 and may explain the widely divergent delta FEV1 values reported in the literature. After adjusting for survey biases, the relationships of delta FEV1 to age, sex, and body size were examined in "healthy" nonsmokers in the population who were followed for an average of 9.6 yr. The FEV1 showed an accelerating decline with age, and delta FEV1 was best explained by a model of proportional change in which it is a function of Age X Height3. Longitudinally determined delta FEV1 showed much less decline in function and a later apparent age of onset of decline than suggested by cross-sectional analyses. While small "learning effects" in longitudinal calculations cannot be excluded, most of the cross-sectional versus longitudinal differences are probably ascribable to past events that have adversely affected the FEV1 in older adults, thereby increasing cross-sectional age effects. Present findings also suggest that generally applicable "prediction equations" for delta FEV1 probably cannot be developed and that delta FEV1 values should only be compared between groups studied in an identical manner within a given study.(ABSTRACT TRUNCATED AT 250 WORDS)