Several single-sample, Cr-51 EDTA based methods of estimating glomerular filtration rate in children have been proposed. The authors retrospectively reviewed our experience in 467 pediatric patients using Tc-99m DTPA and 4 ultrafiltered serum samples drawn 2-4 hours postinjection. The authors used the 2-hour specimens for the single-sample method. Regression analysis of the 2 techniques revealed significant scatter. Mathematically, it was shown that the error of the single-sample method was minimized when the clearance t 1/2 equalled the sampling time x 0.693. This model predicted a 300% error when the t 1/2 was varied to its observed extremes. The observed error of the single-sample method fit this theoretical prediction very well. Analysis of a two-sample method versus the four-sample method showed no improvement in results. In a subset of 18 unfiltered plasma samples, the glomerular filtration rate was falsely low in 12 patients. The authors conclude that multiple blood samples and serum ultrafiltration are required to accurately measure glomerular filtration rate when using Tc-99m DTPA.