A new chronic kidney disease epidemiology collaboration (CKD-EPI) equation uses the same four variables as the modification of diet in renal disease (MDRD) equation but appropriate interpretation of estimated glomerular filtration rate (eGFR) when switching to the CKD-EPI equation requires understanding of its performance. The aim of this study is to compare CKD-EPI and MDRD equations in elderly women. All serum creatinine results were extracted from the medical records of the 208 female patients with Alzheimer’s dementia and other dementias according to MKB-10 disease classification. Median (interquartile range) age was 80 (74-84) years. A kinetic compensated Jaffe assay on an Integra 400 analyser was used for creatinine measurement. The assay has been standardized against ID-MS reference method. CKD-EPI eGFR was statistically significantly higher than MDRD eGFR when whole sample was taken into account (Wilcoxon test; Z = -3.299; P = 0.001). Agreement between GFR categories calculated by MDRD and CKD-EPI equations on overall sample was substantial (κ = 0.759). When age and presence of somatic diseases were combined, the largest agreement improvement was found in participants aged ≥70 with somatic diseases (κ = 0.804). In all other groups agreement rate was reduced: in those aged <70 with somatic diseases (κ = 0.731), and in those aged ≥70 with no somatic diseases (κ = 0.737) but still substantial. For the participants aged ≤70 years without somatic disease the agreement was reduced (κ = 0.412). The agreement between GFR categories calculated by MDRD and CKD-EPI equations on overall sample was substantial but CKD-EPI equation produced higher median eGFR then MDRD equation in elderly patients particularly among participants aged ≥70 years at GFR ≥60 ml/min/1.73 m².