Introduction: The aim of this study was to determine which indicator of obesity index is the best predictor of type 2 diabetes mellitus in Iranian women. Materials and Methods: In an urban Iranian population, we examined 2801 females, aged ≥ 20 years who were non-diabetic at baseline. Over a median 3.5 year (11months to 6.3y) follow up, we compared the ability of central obesity indicators [waist circumference (WC), waist-to-hip ratio (WHR), waist-to-height ratio (WHtR)] to body mass index (BMI) in the prediction of type 2 diabetes, based on American Diabetes Association 2003 criteria. To estimate odds ratio (OR) of developing diabetes associated with each quartile of anthropometric variables, we used a multivariate model adjusted for age, hypertension, triglycerides, HDL-C, family history of diabetes (model 1) and a full model adjusted for the variables in model (1) plus abnormal glucose tolerance at the baseline (model 2). Receiver operator characteristic (ROC) analyses were used to determine the predictive power of each indicator in the development of type 2 diabetes, after adjustment for age. Results: During our follow up, 114 individuals developed diabetes (4.1%). The OR (95% CIs) of developing diabetes increased to 4.8 (2.1-10.9), 6.7 (2.6-17.1), 8.7 (3.0-24.7), 8.0 (3.1-20.6) for the highest quartiles of BMI, WC, WHR and WHtR compared to the lowest quartile in model (1) respectively. In model (2) the ORs (95% CIs) of the fourth quartile of anthropometric variables decreased, compared to value in model (1), but remained significant [BMI: 3.1 (1.3-7.2), WC: 3.1 (1.1-8.3), WHR: 4.0 (1.3-11.8), WHtR: 3.3 (1.2-8.8)]. Compared to BMI, the WHtR had a higher area under ROC curve (0.72 vs 0.69 P<0.05), while there was no difference between ROCs of BMI and other anthropometric variables. Conclusion: In Iranian women, WHtR is a better predictor of type 2 diabetes than BMI.