Aims: The Cockcroft-Gault (CG), the Modification of Diet in Renal Disease (MDRD) and the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) formulae are often used to estimate glomerular filtration rate (GFR). The objective was to determine the best method for estimating GFR in older adults. Methods: A cross-sectional study was conducted at the geriatric wards of two hospitals in The Netherlands. Patients aged 70 years or above with an estimated (e)GFR below 60mlmin-11.73m-2 were included. The CG, CG calculated with ideal bodyweight (IBW), MDRD and CKD-EPI formulae were compared with a criterion standard, sinistrin clearance. Renal function was classified into five stages according to the National Kidney Foundation Disease Outcomes Quality Initiative chronic kidney disease classification, as follows (in mlmin-11.73m-2): stage1, eGFR≥90; stage2, eGFR of 60-89; stage3, eGFR of 30-59; stage4, eGFR of 15-29; and stage 5, eGFR<15. Results: Sixteen patients, 50% male, with a mean age of 82 years (range 71-87years) and mean body mass index 26kgm-2 (range 18-36kgm-2), were included. On average, all formulae slightly overestimated GFR, as follows (in mlmin-11.73m-2): CG +0.05 [95% confidence interval (CI) -28 to +28]; CG with IBW +0.03 (95% CI -20 to +20); MDRD +9 (95% CI -16 to +34); and CKD-EPI +5 (95% CI -20 to +29). They classified kidney disease correctly in 68.8% (CG), 75% (CG with IBW), 43.8% (MDRD) and 68.8% (CKD-EPI) of the participants, respectively. Conclusions: The CG, CG with IBW, MDRD and CKD-EPI formulae estimate the mean GFR of a population rather well. In individual cases, all formulae may misclassify kidney disease by one stage.