In this report of a 5-yr longitudinal study of workers employed at six cotton textile mills, exposure and across-shift FEV1 change were evaluated as possible predictors of the annual change in FEV1 for yarn manufacturing workers. A total of 611 workers had three repeatable spirometric tests, over at least 3 yr, and at least one (average of three) across-shift test, while always working the same shift. The "same shift" criterion controlled for the effect of diurnal variation. Average exposure was determined from measures of lint-free elutriated cotton dust in combination with job histories. This study found a significant association between the acute and chronic effects of cotton dust exposure. Both exposure and across-shift change proved to be significant predictors of annual change, and excess annual declines in FEV1 were predicted even for exposures of 200 micrograms/m3 and across-shift drops in FEV1 of 200 ml. These results suggest that, to prevent dust-related chronic decline in lung function, current smokers should be excluded from yarn manufacturing work and exposures should be reduced below 200 micrograms/m3, to approximately 100 micrograms/m3.