AIMS/HYPOTHESIS: It is argued that GFR estimation (eGFR) using cystatin C-based equations (eGFRcys) is superior to that using creatinine-based equations (eGFRcre). We investigated whether eGFRcys are superior to eGFRcre in patients with type 2 diabetes.METHODS: GFR was measured in 448 type 2 diabetic patients using (51)Cr-EDTA-measured GFR (mGFR) as the reference standard. Bias, precision and accuracy of eGFRcys and eGFRcre were compared.RESULTS: The most accurate eGFRcre equation (Chronic Kidney Disease Epidemiology Collaboration [CKD-EPI]), which produced the highest proportion of estimates that were within 30% and 10% of the reference standard (80.7% and 38.0% of samples, respectively) had a bias of 7.1 and precision of 12.0 ml min(-1) 1.73 m(-2). The calibrated eGFRcys with the highest accuracy (Tan-C), which produced the highest proportion of estimates that were within 30% (78.8%) and within 10% (39.0%) of the reference standard had a bias of -3.5 and precision of 18.0 ml min(-1) 1.73 m(-2). Moreover, the areas under the receiver operating curve were higher with eGFRcre (CKD-EPI and Modification of Diet in Renal Disease [MDRD]) than with eGFRcys for the diagnosis of mild (mGFR <90 ml min(-1) 1.73 m(-2)) and moderate (mGFR <60 ml min(-1) 1.73 m(-2)) chronic kidney disease. In patients with mGFR ≥90 ml min(-1) 1.73 m(-2), CKD-EPI was the least biased, the most precise and the most accurate equation.CONCLUSIONS/INTERPRETATION: In patients with type 2 diabetes, eGFRcys do not currently provide better eGFR than eGFRcre. At present, compared with eGFRcys, eGFRcre are better at predicting the stage of chronic kidney disease. In addition, CKD-EPI seems to be the best equation for eGFR in patients with normal renal function.