OBJECTIVE: Epidemiological data on the association between different aspects of adiposity and the risk for chronic kidney disease (CKD) in a cohort are limited. We compared the independent power of waist circumference (WC), waist-to-hip ratio (WHR), and body mass index (BMI) in predicting CKD in a large cohort of adults. DESIGN: This was a population-based cohort study. SETTING, PARTICIPANTS, AND MEASUREMENTS: A representative sample of 3107 subjects (1309 men and 1798 women), aged over 20 years, and free of CKD at baseline, was followed for 7 years. METHODS: We estimated glomerular filtration rate (GFR) by using the abbreviated equation from the Modification of Diet in Renal Disease Study, and defined CKD as an estimated GFR of less than 60 mL/min/1.73 m(2). Adjusted relative hazards of CKD were modeled using Cox proportional hazards regression, including BMI, WC, and WHR as risk factors. RESULTS: During the 7-year follow-up (median of 2183 days), 13.5% of participants (n = 419) developed CKD. The WC was positively related to risk of CKD, after multivariable adjustment for age, sex, smoking, menopause, physical activity, blood pressure, prevalent and incident diabetes, and change in WC during study period: Hazard ratios for CKD incidence were 1.00 (reference), 1.60 (95% CI: 1.06, 2.42), 1.86 (0.95 CI: 1.21, 2.85), and 1.88 (0.95 CI: 1.17, 3.01) for WC categories 1 to 4, respectively (P for trend < .02). The WHR was not independently associated with CKD. The rate of GFR decline (measured in mL/min/1.73 m(2)/year) was associated with baseline waist categories: regression coefficient for 1 SD increase in WC = -0.18 (0.95 CI:-0.28, -0.07). Based on Harrell's measure of concordance statistics, baseline WC was a better predictor of CKD than WHR (P < .05) and BMI (P < .05). CONCLUSIONS: Abdominal adiposity measured with WC, irrespective of general adiposity, is a more important determinant of CKD risk in adults than are WHR and BMI.