The relation of post-load plasma glucose to 12-year cancer mortality was studied in 11,521 white men and 8,591 white women aged 35-64 years at entry in the Chicago Heart Association Detection Project in Industry. There were 298 deaths in which cancer was the underlying cause in men and 186 such deaths in women. Plasma glucose levels at baseline measured one hour after a 50 g oral glucose load were positively related to cancer mortality in men but not in women. This increased mortality in men persisted after controlling for age, body mass index, smoking, serum cholesterol, systolic blood pressure, education, and anti-hypertensive treatment. A site-specific analysis of the male cancer deaths demonstrated higher mean plasma glucose levels for decedents for nearly all sites except oral cancers, liver cancer, and other respiratory cancers. For women, elevations existed for those who died of colon, lung, and hematologic/lymphatic tumors, but not for most other sites. The associations between baseline post-load plasma glucose levels and cancer mortality were similar for earlier and later cancer deaths, which indicates that incipient or occult neoplasm at baseline was not responsible for elevated glucose levels in men who subsequently died of cancer. Cancer mortality rates were also higher for those with a clinical diagnosis of diabetes than for those without, but the numbers of persons with diabetes were small as were the numbers of cancer deaths (19 in the men and 10 in the women).