The determination of renal function is crucial in patients with type 2 diabetes (T2DM), a population at risk for chronic kidney disease (CKD). Glomerular filtration rate (GFR) can be measured (mGFR) with gold standard methods or estimated (eGFR) with formulas. Since 1957, when Effersoe published the first formula, more than 50 equations have been developed to estimate GFR. In this review, we examined the studies that compared mGFR and eGFR in patients with T2DM to analyze the performance of those formulae in this population. In cross-sectional studies, the average error of eGFR was ±30% of mGFR. Thus, in a patient with mGFR of 60 mL/min, eGFR may vary from 42 to 78 mL/min. Moreover, many patients were misclassified according to CKD stages. Formulas failed to detect glomerular hyperfiltration. In longitudinal studies, eGFR poorly reflected real GFR decline over time. All studies showed that eGFR decline was slower than mGFR decline. Notably, no major improvement in accuracy and precision has been observed since 1957 despite the use of cystatin-c. Thus, formulas are not reliable indicators of GFR in patients with T2DM. In clinical studies, where GFR is the main outcome measure of the study, eGFR should be avoided.