Two protocols for the determination of glomerular filtration rate (GFR) from sinistrin clearance are considered: a bolus injection and a bolus followed by infusion. On both occasions, serial blood and urine samplings are scheduled up to 6 h. Four calculation methods are compared for estimating GFR from the data obtained during each protocol: classical UV/P (ratio of urinary excretion rate over plasma concentration) after bolus or bolus plus infusion; 2-point (log-linear slope multiplied by apparent volume of distribution); D/AUC (ratio of dose over area under the curve) after bolus; and Rin/P (ratio of infusion rate over steady-state concentration) during infusion. Some refinements of the calculations are devised. Data are simulated by running a bicompartmental pharmacokinetic model with renal elimination, and contaminating the values with an array of random errors. The statistical performance of the respective calculation methods is assessed by graphical means. The UV/P method performs poorly during 2 hours following the bolus; on both bolus and infusion data, it suffers from imprecision on the urinary volume. The 2-point method is acceptable between 1 and 4 h after bolus; later, the estimates become much less precise. The D/AUC method appears highly reliable when integrating the concentrations up to 3 h after bolus; it requires extrapolation towards infinity. The Rin/P method is satisfactory if applied later than 3 to 4 h after the loading dose. The advantages and drawbacks of each methods must be evaluated in relation with the particular clinical setting in which GFR is to be estimated. D/AUC represents the most advisable approach for snapshot renal testing in subjects or patients without important renal impairment.