Several lines of evidence support the notion that elevated blood viscosity may predispose to insulin resistance and type 2 diabetes mellitus by limiting delivery of glucose, insulin, and oxygen to metabolically active tissues. To test this hypothesis, the authors analyzed longitudinal data on 12,881 initially nondiabetic adults, aged 45–64 years, who were participants in the Atherosclerosis Risk in Communities (ARIC) Study (1987–1998). Whole blood viscosity was estimated by using a validated formula based on hematocrit and total plasma proteins at baseline. At baseline, estimated blood viscosity was independently associated with several features of the metabolic syndrome. In models adjusted simultaneously for known predictors of diabetes, estimated whole blood viscosity and hematocrit predicted incident type 2 diabetes mellitus in a graded fashion (Ptrend (linear) < 0.001): Compared with their counterparts in the lowest quartiles, adults in the highest quartile of blood viscosity (hazard ratio = 1.68, 95% confidence interval: 1.53, 1.84) and hematocrit (hazard ratio = 1.63, 95% confidence interval: 1.49, 1.79) were over 60% more likely to develop diabetes. Therefore, elevated blood viscosity and hematocrit deserve attention as emerging risk factors for insulin resistance and type 2 diabetes mellitus.