Variation among clinical laboratories in calibration of serum creatinine assays is a source of error in glomerular filtration rate (GFR) estimation equations. We evaluated impact of this variation on GFR estimates. Errors in GFR estimates were computed based on the range of calibration differences from the 1994 College of American Pathologists (CAP) survey using the Modification of Diet in Renal Disease (MDRD) Study GFR equation. Mean (95% CI) calibration difference observed in the CAP survey was +0.14 mg/dL (+12.4 micromol/L) [-0.09, +0.37 mg/dL (-7.96, +32.71 micromol/L)]. Errors in GFR estimates using uncalibrated serum creatinine values were lower in individuals with lower estimated GFR. For GFR of 60 mL/min/1.73 m(2), the mean calibration difference in the CAP survey was associated with errors in GFR estimation between -5.5 to -8.1 mL/min/1.73 m(2) (-9.1 to -13.5%) depending on race and sex. The 95% confidence interval for the calibration difference was associated with a maximal range of error in GFR estimates from +4.6 to -18.1 mL/min/1.73 m(2) (+7.6 to -30.2%). Errors of this magnitude at an estimated GFR of 60 mL/min/1.73 m(2) are not likely to be of clinical significance. However, errors at higher levels of estimated GFR would be greater, making GFR estimates in this range unreliable. Recalibration of serum creatinine assays to the MDRD Study clinical laboratory would improve accuracy of GFR estimation using the MDRD Study equation, but is not practical for all clinical laboratories. As an interim solution, clinical laboratories could report GFR estimates <60 mL/min/1.73 m(2) without recalibration with an acceptable accuracy.