The precision and reporducibility of three different clearance methods as used in clinical routine assessment of glomerular filtration rate (GFR) were investigated in 51 patients: total [51Cr]EDTA plasma clearance (E); 24-hr endogenous creatinine clearance (C); and creatinine clearance estimated from the plasma creatinine concentration, weight, and sex- and age-dependent mean creatinine excretion rate (c). The precision and reporducibility (coefficient of variation) for single determinations were, in patients with E greater than 30 ml/min, 5.5 and 4.1% (E); 26.9% (C); and 23.2 and 11.0% (c). The corresponding figures for E less than 30 ml/min were 11.6 and 11.5% (E); 21.9% (C); and 21.4 and 6.5% (c). The precision of C could not be ameliorated by excluding single deviating determinations, but only by excluding patients for whom the precision of 15.5% for mean of three determinations of C (total material) could be reduced to 10% by excluding 25% of the patients. The present data indicate that E in most cases is the method of choice for assessment of GFR in clinical routine work. For changes in renal function, especially at low functional levels, c may be of value.