We evaluated associations between dust exposure, demographic factors, and lung function by longitudinal and cross-sectional analyses in 475 steelworkers who participated in at least three spirometry tests over 5 yr between 1982 and 1991. Baseline and follow-up spirometry and changes between baseline and final follow-up assessment attributable to age, height, weight, weight gain, smoking status, pack-years, and years worked in dusty areas were examined using stepwise multiple linear regression techniques. Smoking, aging, being overweight, excessive weight gain, and dust exposure were related to a lower level and a steeper slope of decline of pulmonary function. Cigarette smoking was also an important risk factor. Dust exposure was related to the level of lung function, with a stronger effect at baseline than at follow-up. Estimated loss at baseline of FEV1, FVC, and FEV1/FVC% was 9.3, 6.4 ml, and 0.1 % per year of employment in a dusty area, respectively, whereas the association between dust exposure and longitudinal decline of lung function was weak. However, a strong relationship between weight gain and longitudinal decline of FEV1 and FVC was found. Estimated decreases in FEV1 and FVC attributable to weight gain were 4.7 and 6.3 ml per lb/yr, respectively. This work suggests that weight gain is an important determinant for longitudinal lung function decline. This large impact of weight gain in the decline of lung function in a middle-age and relatively overweight working population has not been previously reported. Additional work needs to be undertaken to show the strength of this relationship in other populations.