Chronic kidney disease (CKD) is a progressive disease leading to loss of glomerular filtration rate (ΔGFR, measured in ml/min/1.73 m 2/year). ΔGFR is usually assumed to be constant, but the hyperfiltration theory suggests that it accelerates in severe uraemia. A retrospective analysis of estimated GFR (eGFR) calculated from the Modification of Diet in Renal Disease equation was performed to evaluate whether ΔGFR is constant or accelerating. Methods: 1,441 patients attending a nephrology clinic over a 21-year period, with an initial eGFR <60 ml/min/1.73 m 2 and an observation period ≥2 years, were included. eGFR was calculated from all creatinine measurements. 420 patients developed end-stage renal disease (ESRD). First- and second-order polynomial regression analysis of eGFR against time was performed for each patient individually. Patients had accelerating uraemia progression if the second-order term coefficient was negative. Results: The initial eGFR was 30.8 ±15.1 ml/min/1.73 m 2. The second-order coefficient was median -0.15 ml/min/1.73 m 2/year 2 (interquartile range -0.92, +0.34). Significantly more patients had an accelerating loss (62%, p < 0.001). Acceleration was mainly seen when eGFR was <30 ml/min. ΔGFR was mean 1.47 ± 4.5 ml/min/1.73 m 2/year (male 1.67, female 1.22). ESRD patients lost 5.4 ± 5.4 ml/min/year/1.73 m 2 during the last year before ESRD. Accelerating loss was seen for all diagnoses except polycystic disease. Diagnoses with higher ΔGFR were polycystic renal disease (3.3 ml/min/1.73 m 2/year), hypertensive nephropathy (2.1 ml/min/1.73 m 2/year) and diabetic nephropathy (2.6 ml/min/ 1.73 m 2/year). There was no evidence of improvement in overall uraemia progression during the period of observation. Conclusions: Uraemia progression in CKD stages 3-5 is not linear, but shows an accelerating trend. This suggests that hyperfiltration mechanisms play a role in CKD progression. ESRD cannot thus be predicted from previous ΔGFR alone.