Serum creatinine and endogenous creatinine clearance (CrCl) are widely used measures of renal function. This study compares the precision, bias, and sources of error in using different CrCl measures to estimate the glomerular filtration rate (GFR) in 118 men and women screened for the African-American Study of Kidney Disease and Hypertension (AASK) pilot study. We measured serum creatinine, 24-hour CrCl, and CrCl during timed clearance periods conducted simultaneously with an 125I-iothalamate GFR study. Serum creatinine was measured using two different kinetic rate Jaffe methods (CX3 and Hitachi). After standardization for body surface area, the different measures of renal function available for each individual were compared with the 125I-iothalamate GFR simultaneous to the CrCl. In a subset of 50 participants, the CrCl measures were compared with a follow-up GFR (fGFR). The mean 125I-iothalamate GFR was 65.2 (SD, 26.4), with a range of 11 to 122 mL/min/1.73 m2. The mean +/- SD percentage differences from the GFR were -9%+/-22% for the Cockcroft-Gault estimated CrCl, 1%+/-29% for the 24-hour CrCl, and 8%+/-16% for the CX3 simultaneous CrCl. The Hitachi method overestimated serum creatinine and underestimated GFR. Compared with an fGFR, the mean +/- SD differences were 2%+/-19% for the first GFR, -6%+/-20% for the Cockcroft-Gault estimated CrCl, 10%+/-28% for the 24-hour CrCl, and 14%+/-29% for the CX3 simultaneous CrCl. Thus, the increased precision with which the timed CrCl predicted its simultaneous GFR did not extend to improved ability to predict a future GFR. The fractional excretion of creatinine, measured as the ratio of the CX3 simultaneous CrCl to 125I-iothalamate clearance, increased with decreasing GFR but was lower than expected (mean +/- SD of 1.21+/-0.16 for GFRs between 20 and 40 mL/min/1.73 m2). The lower fractional excretion explains why the 24-hour and Cockcroft-Gault CrCls did not overestimate GFR, but the reasons for this lower excretion are uncertain. Creatinine assay specificity and calibration are important sources of variability that must be examined in any CrCl measure of GFR. We conclude that despite requiring substantially more time and effort, neither the outpatient 24-hour urine nor the timed CrCl offered increased precision over a calculation based on serum creatinine, sex, age, and weight in predicting GFR.