The present investigation represents an update of a previous cohort mortality study of 7543 workers who were employed at a petroleum refinery in Beaumont, Texas, for at least 1 year between 1945 and 1996. The updated study covered an observation period of 51 years, from 1946 to 1996, with a total of 208,627 person-years of observation. A total of 3020 (40.0%) cohort members were known to have died. The mortality data were analyzed in terms of cause-specific standardized mortality ratios (SMRs) and 95% confidence intervals (95% CIs). The overall mortality of the cohort was significantly lower than expected when compared with that of the general US population (SMR, 95.7; 95% CI, 92.3 to 99.2). Overall cancer mortality was also lower than expected (SMR, 85.8; 95% CI, 79.4 to 92.5). For specific cancer sites, significant mortality deficits were observed for the following: buccal cavity and pharynx, esophagus, large intestine, rectum, larynx, lung, and bladder and other urinary organs. No significant increase was reported for any site-specific cancer. A non-significant increase in acute myeloid leukemia was observed among male employees (SMR, 147.2; 95% CI, 76.1 to 257.2). Detailed analyses indicated that the excess was restricted to workers hired before 1950. No increase was detected for other leukemia cell-types, non-Hodgkin’s lymphoma, or multiple myeloma. For non-malignant diseases, the majority of SMRs were below 100, and no significant increase was observed for any cause. In particular, significant mortality deficits were reported for ischemic heart disease (SMR, 91.0; 95% CI, 85.4 to 96.9), non-malignant respiratory disease (SMR, 61.5; 95% CI, 52.2 to 72.0), pulmonary fibrosis (SMR, 51.0; 95% CI, 22.0 to 100.4), cirrhosis of the liver (SMR, 47.2; 95% CI, 30.6 to 69.7), and accidents (SMR, 81.7; 95% CI, 66.3 to 99.6). Separate analyses of male workers by job classification (process and maintenance) were conducted. Mortality from acute myeloid leukemia was elevated among employees in maintenance jobs (8 observed deaths vs 4.31 expected; SMR, 185.5; 95% CI, 80.1 to 365.6). However, no upward trend by length of service was found. A detailed analysis indicated that the acute myeloid leukemia mortality excess was limited to maintenance workers who were hired before 1950. No other significant excess was detected for any cause among maintenance or process workers. These findings from the present study were discussed in conjunction with results from previous investigations of employees at the Beaumont refinery and with results from other refinery studies. Potential limitations of the study were also discussed.