We extended a previous cohort study of workers at a metal components manufacturing facility, which found elevated rats of lung cancer and unspecified nonmalignant respiratory diseases, by evaluating mortality rates from 1950 to 1987 for 3,630 workers categorized according to potential exposures to metal dusts, cutting oils/fluids, metal fumes, and solvents. Lung cancer rates were elevated for workers with exposure to metal dusts and cutting oils/fluids [standardized mortality ratio (SMR) = 1.8; 95% confidence interval (CI) = 1.2-2.6]. We found that elevated mortality was restricted to those workers first exposed during 1950-1959 (SMR = 2.5; 95% CI = 1.6-3.7) and that mortality rates were similar for workers with exposure durations of < 2 years, 2-9 years, and > or = 10 years, respectively. The elevated mortality for workers with < 2 years exposure and the lack of a trend with exposure duration raise the possibilities of either a short-lived occupational carcinogen or an unspecified confounding factor. Mortality rates for stomach or pancreatic cancers--cancers frequently reported to be in excess for metalworking populations--or for nonmalignant respiratory diseases were consistent with expected values for all exposure subgroups.