Low socioeconomic status is associated with high mortality, but the extent to which socioeconomic status affects particular diseases and whether socioeconomic status effects have changed over time are uncertain. The authors used education as a marker for socioeconomic status in a study of two large American Cancer Society cohorts (follow-up, 1959-1996). Low education was associated with higher death rates in both cohorts from all causes and most specific causes, except breast cancer and external causes among women. Life expectancy in the more recent cohort was 4.8 years shorter for men and 2.7 years shorter for women for the least versus the most educated. The inverse relation between education and mortality was strongest for coronary heart disease, lung cancer, diabetes, and chronic obstructive pulmonary disease; moderate for colorectal cancer, external causes (men only), and stroke; weak for prostate cancer; and reversed for external causes among women. The direction of a weak gradient for breast cancer differed for those with and without prevalent breast cancer at baseline. Adjustment for conventional risk factors, probable intermediate variables between education and mortality, diminished but did not eliminate the observed educational/mortality gradients. Temporal trends showed increasing mortality differences by education for coronary heart disease, diabetes, and lung cancer for women.