BACKGROUND In 1997, the International Agency for Research on Cancer classified 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) as a group 1 human carcinogen, based largely on four highly exposed industrial cohorts that showed an excess of all cancers combined. In this study, we extended the follow-up period for the largest of these cohorts by 6 years and developed a job-exposure matrix. METHODS We did cohort mortality analyses involving 5132 chemical workers at 12 U.S. plants by use of life table techniques (U.S. population referent) and Cox regression (internal referent). We conducted exposure-response analyses for 69% of the cohort with adequate work history data and adequate plant data on TCDD contamination. All P values are two-sided. RESULTS The standardized mortality ratio (SMR) for all cancers combined was 1.13 (95% confidence interval = 1.02-1.25). We found statistically significant positive linear trends in SMRs with increasing exposure for all cancers combined and for lung cancer. The SMR for all cancers combined for the highest exposure group was 1.60 (95% confidence interval = 1.15-1.82). SMRs for heart disease showed a weak increasing trend with higher exposure (P = .14). Diabetes (any mention on the death certificate) showed a negative exposure-response trend. Internal analyses with Cox regression found statistically significant trends for cancer (15-year lag time) and heart disease (no lag). CONCLUSIONS Our analyses suggest that high TCDD exposure results in an excess of all cancers combined, without any marked specificity. However, excess cancer was limited to the highest exposed workers, with exposures that were likely to have been 100-1000 times higher than those experienced by the general population and similar to the TCDD levels used in animal studies.