Background: Obesity is a strong risk factor for hyperglycemia. However, the performance of waist-to-height ra- tio (WHtR), waist-to-hip ratio (WHR), waist circumference (WC) and body mass index (BMI) in predicting hy- perglycemia was inconclusive; and ethnic differences may exist in the associations. Objective: The objective was to compare WHtR, WC, WHR and BMI in predicting hyperglycemia among Chinese adults and identify optimal cut-off points. Design: A community-based prospective cohort study was conducted during 2004-2007 in Nan- jing China. WHtR, WC, WHR, BMI, fasting capillary blood glucose, covariates and potential confounders were assessed at baseline and third year of follow-up. Results: The overall cumulative incidence of hyperglycemia was 8.6% (8.0% in men, 9.0% in women). Relative risks across quartiles of WHtR, WHR, WC and BMI were 1.00, 1.33, 1.67, 3.40; 1.00, 2.38, 2.53, 3.87; 1.00, 1.29, 1.75, 2.90; and 1.00, 1.45, 1.49, 2.41 in men, and 1.00, 1.24, 1.99, 2.87; 1.00, 1.14, 2.28, 2.66; 1.00, 1.32, 1.80, 3.14; and 1.00, 1.39, 1.50, 2.08 in women, respectively. p for trend was <0.01 for each marker and gender. Adjustment for potential confounders did not change such dose-response relationships materially. ROC analysis indicated that WHtR had the best sum of sensitivity and specificity compared to the other measures .Optimal cut-offs for WHtR, WHR, WC and BMI were 0.51, 0.92, 85 and 24 for men, while 0.55, 0.86, 82 and 25 for women, respectively. Conclusion: WHtR, WHR, WC and BMI were positively associated with subsequent hyperglycemia. WHtR and WC best predicted hyperglycemia among Chinese adults.