The association of nonfasting blood glucose levels with CVD incidence was determined prospectively in 1382 men and 2094 women aged 45 to 84 years participating in the Framingham Heart Study. For this study, all patients were classified in 1970 as diabetic or nondiabetic. Every 2 years they were examined, categorized according to causal blood glucose samples obtained at the clinic visit, reclassified for development of CVD and diabetes mellitus, and followed 10 years for CVD. During the follow-up period, 350 men and 369 women developed CVD. Age-adjusted CVD rates were positively associated with glucose levels in nondiabetic women who did not develop diabetes during follow-up. No such associations were seen in men. Multivariate analyses confirmed the independent association of blood glucose levels with later CVD in nondiabetic women. This study shows that hyperglycemia in the original Framingham cohort is an independent risk factor for CVD in nondiabetic women, but not among men.