The aim of this study was to develop a method to predict the glomerular filtration rate (GFR) in children by using the population pharmacokinetic approach. This powerful approach is widely used for drug development in order to study relationships between patients' characteristics (demographic, morphological, biological covariates) and pharmacokinetic parameters. For the first time, (51)Cr-EDTA plasma concentrations from 64 children (development data set) were analyzed using the Non-linear Mixed Effects Model (NONMEM) program to determine the most appropriate equation to relate (51)Cr-EDTA clearance (as a measurement of GFR) and patient characteristics. The most predictive equation was based on body weight, square height, and plasma creatinine (PCr, determined by the Jaffé method). This equation was then validated using the data from a further 33 patients. This equation produced estimates of GFR that were less biased and more precise than those obtained using the widely used Schwartz formula. The coefficient of correlation between estimated and actual GFR was 0.83, and the 10th to 90th percentiles for percentage errors were -20% to +30%. Finally, analysis of the whole data set (97 patients) led to an equation (i.e., GFR (ml/min)=[56.7 x Body weight (kg)+0.142 x Length(2)(cm)]/PCr ( microM)) very similar to that obtained from the development data set. This equation would be useful for estimating GFR in children when isotopic determination of the (51)Cr-EDTA clearance cannot be performed.