A previous cross-sectional analysis of 1980 data from a population of working men in the Paris area has shown a significant relationship of blood eosinophilia to a reduced FEV1 among nonsmokers, remaining after excluding men with a history of asthma. In the present report, we reexamine this relationship, after taking into account asthma, bronchial hyperresponsiveness, and positive skin prick tests, using data collected in 1985 in a subsample of 363 men from the initial population. Blood eosinophilia, defined by 5% or more eosinophils or by 250 or more eosinophils per cubic millimeter appeared to be associated with a lower FEV1, primarily in nonsmokers. A difference of approximately 0.40 L was observed in never-smokers with eosinophilia (greater than or equal to 5% of eosinophils) compared with those without. This association persisted after exclusion of subjects with atopy, asthma, and bronchial hyperresponsiveness. Longitudinally, no significant association was observed between 1980 eosinophilia and the annual FEV1 decline between 1980 and 1985, even in nonsmokers. The results of our cross-sectional analyses suggest that asthma or asthma-like disorder does not explain the association between eosinophilia and FEV1. The role of eosinophil in respiratory disorders may go beyond its intervention in allergy. Further longitudinal studies are needed to better understand discrepancies between cross-sectional and longitudinal data and whether eosinophilia is a risk factor for chronic air-flow limitation.