To assess the performance of the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation at baseline and longitudinally in people with type 2 diabetes.Adults with type 2 diabetes attending Austin Health, Melbourne, with≥3 prospective GFR measurements were included in this retrospective study. Plasma disappearance rate of DTPA (diethylene-triamine-penta-acetic acid) was used to calculate measured GFR (mGFR) and compared to estimated GFR (eGFR). The agreement between mGFR and eGFR was estimated using Intraclass Correlation Coefficient (ICC).152 patients had a median of 4 (IQR: 3, 5) mGFR measurements over a period of 11years (IQR: 9, 12). The difference between mGFR and eGFR increased proportionally to the magnitude of the GFR, increasing by 0.2ml/min/1.73m2 for every 1ml/min/1.73m2 increase in mGFR, indicative of proportional bias. At lower mGFR levels, eGFR overestimated mGFR, and at higher mGFR levels, eGFR underestimated mGFR. There was a significant association between LDL cholesterol, triglycerides, HbA1c, diastolic blood pressure and the difference between mGFR and eGFR.The CKD-EPI formula underestimates mGFR and the rate of decline of mGFR in patients with type 2 diabetes with an mGFR greater than 60ml/min/1.73m2. The association between LDL cholesterol, triglycerides, HbA1c, diastolic blood pressure and the difference between mGFR and eGFR warrants further study.