Aims: We sought to examine whether waist-to-thigh ratio (WTR) performed better than waist-to-height ratio (WHtR), waist-to-hip ratio (WHpR), waist circumference (WC), or body mass index (BMI) in relation to diabetes among US adults. Methods: Data of 6277 men and nonpregnant women 20 years or older from the Third National Health and Nutrition Examination Survey (1988-1994) were analyzed. Results: In men, AUC of WTR (0.83) was larger than that of WHtR (0.78) (P = 0.003), WHpR (0.79) (P < 0.001), WC (0.76) (P < 0.001), and BMI (0.72) (P < 0.001) for diabetes. In women, the AUC of WTR (0.80) was similar to that of WHtR (0.80) (P = 0.89), WHpR (0.79) (P = 0.55), and WC (0.78) (P = 0.36), but larger than that of BMI (0.73) (P = 0.03) for diabetes. After adjustment for potential confounders, WTR had the strongest association with diabetes in men (OR, 2.13; 95% CI, 1.57-2.88; per 1 SD increment), whereas WHpR had the strongest association with diabetes in women (OR, 1.94; 95% CI, 1.60-2.35). Conclusions: WTR performed better than other four indices in men and WTR performed similarly to WHtR, WHpR, and waist circumference, but better than BMI in women for the association with diabetes. Published by Elsevier Ireland Ltd.