OBJECTIVE To assess the association between high but nondiabetic blood glucose levels and the risk of death from all causes, coronary heart disease (CHD), cardiovascular disease, and neoplasms. METHODS We studied the 20-year mortality of non-diabetic, working men, age 44-55 years, in three European cohorts known as the Whitehall Study (n = 10,025), the Paris Prospective Study (n = 6,629), and the Helsinki Policeman Study (n = 631). These men were identified by their 2-h glucose levels following an oral glucose tolerance test and by the absence of a prior diagnosis of diabetes. As the protocol for the oral glucose tolerance test and methods for measuring glucose differed between studies, mortality was analyzed according to the percentiles of the 2-h and fasting glucose distributions, using the Cox's proportional hazards model. RESULTS Men in the upper 20% of the 2-h glucose distributions and those in the upper 2.5% for fasting glucose had a significantly higher risk of all-cause mortality in comparison with men in the lower 80% of these distributions, with age-adjusted hazard ratios of 1.6 (95% CI 1.4-1.9) and 2.0 (1.6-2.6) for the upper 2.5%. For death from cardiovascular and CHD, men in the upper 2.5% of the 2-h and fasting glucose distributions were at higher risk, with age-adjusted hazard ratios for CHD of 1.8 (1.4-2.4) and 2.7 (1.7-4.4), respectively. CONCLUSIONS If early intervention aimed at lowering blood glucose concentrations can be shown to reduce mortality, it may be justified to lower the levels of both 2-h and fasting glucose, which define diabetes.