The aim of this study is to determine which indicator of chronic kidney disease most closely correlates with 10-year Framingham coronary heart disease (CHD) risk among serum creatinine, serum cystatin C (S-CysC), urine albumin-creatinine ratio (UACR), estimated creatinine-based GFRs (eGFRcre), and estimated CysC-based GFRs (eGFRcys) in patients with obesity and diabetes. Serum creatinine, S-CysC, UACR, and cardio-ankle vascular index (CAVI) were examined in 468 outpatients with obesity and type 2 diabetes, free of severe renal dysfunction or previous history of cardiovascular disease, as a cross-sectional survey using baseline data from the multi-centered Japan Diabetes and Obesity Study. S-CysC and eGFRcys had significantly stronger correlations with the 10-year Framingham CHD risk than serum creatinine, eGFRcre, and UACR (creatinine, ρ = 0.318; S-CysC, ρ = 0.497; UACR, ρ = 0.174; eGFRcre, ρ = -0.291; eGFRcys, ρ = -0.521; P < 0.01 by Fisher's z-test). S-CysC and eGFRcys had significantly stronger correlations with CAVI than serum creatinine, eGFRcre, and UACR (creatinine, ρ = 0.198; S-CysC, ρ = 0.383; UACR, ρ = 0.183; eGFRcre, ρ = -0.302; eGFRcys, ρ = -0.444; P < 0.05 by Fisher's z-test). The receiver operating characteristic curves to distinguish the high-risk patients for CHD revealed significantly larger areas under the curve of S-CysC and eGFRcys than those of serum creatinine, UACR, and eGFRcre (serum creatinine, 0.64; S-CysC, 0.75; UACR, 0.56; eGFRcre, 0.63; eGFRcys, 0.76; P < 0.01). The data suggested that eGFRcys can be more predictive of the 10-year CHD risk than eGFRcre in Japanese patients with obesity and diabetes.