A nested case-control study was undertaken to identify the determinants of lung cancer mortality in a cohort of 8147 foundry men among whom an excess of lung cancer deaths was previously observed. The present study consisted of all lung cancer deaths (N = 220) that occurred within this cohort between 1950 and 1989. both living and dead controls, matched on race and attained age, were selected in the ratio of 10:1 (N = 2200) by means of the incidence density sampling procedure. All cases and two controls per case, randomly selected from each case's 10 controls, were included in a smoking history survey. Basic smoking history information was obtained for about 71% of these study subjects. For the purpose of this study, formaldehyde exposure levels were categorized as high, medium, low, and none. Airborne silica exposure was categorized only as high, medium, and low levels, because all foundry workers were known to be exposed to silica. Conditional logistic regression analyses indicated that cigarette smoking was a strong predictor of lung cancer mortality in this cohort. Neither exposure to formaldehyde nor silica exposure level, nor employment in any of the six major work areas within the foundry, showed an association with lung cancer.