The relation of fasting and 2-hour serum insulin to the risk for fatal cardiovascular disease was examined in men and women without diabetes. Between 1984 and 1987, 80% of all surviving local members of the Rancho Bernardo Study cohort had measures of insulin and glucose levels obtained before and after a 75-g oral glucose tolerance test. Over the next 5 years, there were 24 cardiovascular disease deaths among 538 men and 21 cardiovascular disease deaths among 705 women. Fasting insulin was unrelated to cardiovascular disease death in men or women; 2-hour insulin was significantly lower in men (but not in women) who died from cardiovascular disease. In men, a 1-standard deviation increase in 2-hour insulin was associated with a 36% reduction in cardiovascular disease mortality (p = 0.01). The significant inverse association of 2-hour insulin with cardiovascular disease death persisted in multiply adjusted models (relative hazard = 0.68; 95% confidence interval 0.47-0.96). Patterns were similar when the analysis was repeated, including men with non-insulin-dependent diabetes mellitus or heart disease at baseline. These findings were not explained by antihypertensive drug use or cigarette smoking. Hyperinsulinemia was not a risk factor for cardiovascular disease in these older men or women. The role of insulin as a cardiovascular disease risk factor requires further investigation.