Data collected in two large-scale longitudinal respiratory studies were analyzed to estimate age-related changes in FEV1 and to determine the between-study differences in change in FEV1 among nonsmoking participants. Two adult cohorts in the Netherlands were followed up for 21 years at 3-year intervals and a cohort of adults in 6 cities in the eastern United States was followed up for 6 years. The age distributions and follow-up rate differed for the two studies. The mean rates of decline in FEV1 for 50-year old male and female never smokers were 27.9 +/- 2.8 mL/yr and 22.1 +/- 0.9 mL/yr, respectively, in the Netherlands cohort. The estimates for men and women in the Six Cities cohort were 33% and 32% greater, respectively, than those for the Netherlands cohort (p < 0.01). The estimated rate of decline in FEV1 associated with each 1-year increase in age was 0.50 +/- 0.12 mL/yr for men and 0.61 +/- 0.08 mL/yr for women in the Netherlands cohort. This estimate was comparable for the Six Cities women and the Netherlands women but significantly different for the two cohorts of men. Inclusion of survey (period) effect in the regression model had only a trivial influence on accelerated rate of decline in FEV1 and did not alter the more rapid acceleration of decline in FEV1 among the Six Cities men. The present study also showed that use of a "prediction equation" for age adjustment to compare across population samples may produce a biased conclusion if the estimated rate of decline in FEV1 associated with age is different and the age distributions are different for the two populations. In addition, height adjustment for FEV1 was found to have a negligible influence on the longitudinal estimates of FEV1 decline in both study samples. In conclusion, substantial between study variation in decline in FEV1 occurs despite use of similar data analysis techniques. Differences in the initial age distributions and loss to follow-up appeared to be important factors in this analysis.