An industry-wide retrospective cohort mortality study was conducted on 6,152 chemical workers (2,460 exposed and 3,692 nonexposed) engaged in chloromethyl ether manufacture at 7 major U.S. companies between 1948 and 1980. A previous study at 6 companies from 1948 through 1972 reported excess respiratory cancer (RC) mortality and significant exposure-response relationships in exposed workers at 1 company (company 2). The present study, which extended follow-up of an additional 7 years for companies 1-6 and included company 7 for follow-up from 1953 through 1980, found excess RC mortality in exposed workers at company 2 [observed (Obs) = 32, standardized mortality ratio (SMR) = 430] and company 7 (Obs = 9, SMR = 603). External comparisons of RC mortality at both companies showed significant exposure-response relationships with respect to cumulative time-weighted exposure. At company 2, where the greatest number of RC deaths occurred, external comparisons showed that RC risk remained constant in relation to age at first exposure and decreased with increasing time since last exposure. With the use of Mantel-Haenszel and relative risk (RR) regression methods, internal comparisons at company 2 demonstrated significant findings of increasing RR with cumulative duration of exposure and cumulative time-weighted exposure and with decreasing time since last exposure. No association was found between RR and age at first exposure. An interesting finding was a significant negative interaction between cumulative time-weighted exposure and age at risk. The best-fitting logistic regression model for the exposed group predicted RR at 2.79 (95% confidence interval = 1.66-4.69) for workers with the mean cumulative exposure score of the 32 RC deaths (lagged by 6 yr) compared with those with negligible exposure (assuming mean age at risk of the RC deaths, 51 years old, and time since last exposure held constant). Qualitative assessment of the results suggests that chloromethyl ether exposure affects both an early as well as a late stage of a putative multistage respiratory malignant process.