Objective: To quantify the relative contribution of elevated 2-hr glucose, fasting glucose (FPG), and HbA1c to all-cause mortality. Study Design and Setting: A joint analysis of two prospective studies with baseline glycemia measurements. Results: The multivariate adjusted hazard ratios (HRs) corresponding to a one standard deviation increase in HbA1c were 1.14 (95% CI 1.03-1.25), 1.08 (0.98-1.19) for FPG and 1.15 (1.05-1.27) for 2-hr glucose, respectively. Entering the 2-hr glucose to the model based on the FPG and HbA1c significantly improved the prediction of mortality, whereas neither FPG, nor HbA1c added significant information once 2-hr glucose was in the models. In subjects with FPG <7.0 mmol/L and HbA1c less than or equal to6.5%, the HR was 1.35 (1.03-1.78) in men with 2-hr glucose greater than or equal to7.8 mmol/L compared with men with 2-hr glucose <7.8 mmol/L. Conclusion: Elevated 2-hr glucose was a predictor of mortality independent of the levels of fasting glucose and HbA1c. (C) 2004 Elsevier Inc. All rights reserved.