Telomeres play a central role in cellular aging, and shorter telomere length has been associated with age-related disorders including diabetes. However, a causal link between telomere shortening and diabetes risk has not been established. In a well-characterized longitudinal cohort of American Indians participating in the Strong Heart Family Study, we examined whether leukocyte telomere length (LTL) at baseline predicts incident diabetes independent of known diabetes risk factors. Among 2,328 participants free of diabetes at baseline, 292 subjects developed diabetes during an average 5.5 years of follow-up. Compared with subjects in the highest quartile (longest) of LTL, those in the lowest quartile (shortest) had an almost twofold increased risk of incident diabetes (hazard ratio [HR] 1.83 [95% CI 1.26-2.66]), whereas the risk for those in the second (HR 0.87 [95% CI 0.59-1.29]) and the third (HR 0.95 [95% CI 0.65-1.38]) quartiles was statistically nonsignificant. These findings suggest a nonlinear association between LTL and incident diabetes and indicate that LTL could serve as a predictive marker for diabetes development in American Indians, who suffer from disproportionately high rates of diabetes.