Longitudinal spirometric data on adults in a general population sample confirmed an overall relationship between rate of change in FEV1 (delta FEV1) and mean FEV1/Ht3. This relationship between FEV1 "slope" and "level" has been called a "horse-racing effect" and is the basis for the widely accepted concept that detecting a low FEV1 will predict a rapid decline in FEV1 and the development of clinically significant COPD. However, when both age and smoking habits were taken into account, the effect persisted only in male smokers. Even in them, it was dependent on the inclusion of part of the delta FEV1 in calculation of the mean FEV1 value. While neither a low initial FEV1/Ht3 nor percent predicted FEV1 significantly predicted a subsequent rapid fall in FEV1, presumably because of regression toward the mean, other initial spirometric variables proved better predictors, at least among male smokers. In them, a low FEV1/FVC ratio on entry was almost uniformly associated with a high rate of decline in FEV1. This was not seen in ex-smokers with similar initial spirometric findings. Also, delta FEV1 could not be predicted with any precision in women regardless of their smoking status. The findings strongly support the concept that detection of early airway disease by spirometric screening will identify those male smokers who are likely to develop clinically significant illness if they continue to smoke and that smoking cessation, even at this point, can markedly improve the prognosis.(ABSTRACT TRUNCATED AT 250 WORDS)