BACKGROUND: Greater central adiposity is related to the risk of diabetes. OBJECTIVE: We aimed to test the hypothesis that central adiposity measured by computed tomography (CT) is a better predictor of the risk of diabetes than is body mass index (BMI), waist circumference (WC), waist/hip ratio (WHR), or waist/height ratio. DESIGN: Visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) were measured at the L2-3 and L4-5 disc spaces in 1106 of the 3234 participants in the Diabetes Prevention Program. Sex-specific proportional hazards models were used to evaluate the association between VAT and SAT at both cuts, BMI, and other measures of central adiposity as predictors of the development of diabetes. RESULTS: Men had more VAT than did women. White subjects had more VAT at both cuts than did other ethnic groups. The ratio of VAT to SAT was lowest in African Americans of both sexes. Among men in the placebo group, VAT at both cuts, WC, BMI, waist/height ratio, and WHR predicted diabetes (hazard ratio: 1.79-1.44 per 1 SD of variable). Among women in the lifestyle group, VAT at both cuts predicted diabetes as well as did BMI, and L2-3 was a significantly better predictor than was WC or WHR. SAT did not predict diabetes. None of the body fat measurements predicted diabetes in the metformin group. CONCLUSIONS: In the placebo and lifestyle groups, VAT at both cuts, WHR, and WC predicted diabetes. No measure predicted diabetes in the metformin group. CT provided no important advantage over these simple measures. SAT did not predict diabetes.