In the final phase of the mortality study of workers at an automotive iron foundry, a subset (N = 3929) of the original cohort of 8147 men, consisting of those exposed to formaldehyde during the period from January 1960 through May 1987, was analyzed. In addition to the external US population, an internal population (N = 2032), consisting of men who had worked in the same foundry during the same time period but not in formaldehyde-exposed jobs, was also used as a referent. Follow-up continued through December 31, 1989. Smoking status was ascertained for 65.4% of the exposed and for 55.1% of the unexposed cohorts. Detailed work histories and evaluation of occupational exposures by an industrial hygienist enabled us to categorize cumulative formaldehyde and silica exposures. Standardized mortality ratios were used to compare the mortality experience of the exposed cohort with the US population and, because of concerns about the healthy worker effect, with an occupational referent population. Relative risks for race, formaldehyde exposure status, smoking status, and silica exposure level were estimated by fitting a Poisson regression model to four causes of death: cancers of the buccal cavity and pharynx, lung cancer, diseases of the respiratory system, and emphysema. No association between formaldehyde exposure and deaths from malignant or nonmalignant diseases of the respiratory system was found. Cigarette smoking and silica exposure were found to be significantly associated with deaths attributed to lung cancer and disease of the respiratory system.