The aim of this study is to develop a population pharmacokinetic model for total and free mycophenolic acid (MPA) in adult kidney transplant patients with chronic graft dysfunction to understand the contribution of potentially relevant covariates to the inter- and the intraindividual variability of MPA in these patients. Twenty patients received mycophenolate mofetil orally up to 1 g twice daily were enrolled in this prospective pharmacokinetic study. Two hundred twenty-nine total and 257 free concentration-time points were determined using reversed-phase high-performance liquid chromatography/ultraviolet at 21 days and 6 months after initiation of mycophenolate mofetil therapy. Data were analyzed using nonlinear mixed effects modeling. A two-compartment model, with zero-order input and first-order elimination from the central compartment, which combined total and free concentration, was selected. Intersubject variability (ISV) was estimated on the clearance, central volume of distribution, and intercompartmental clearance. Only body weight (WT, kg), occasion, and albumin concentration (mg/L) were kept as covariates in the final model. The population pharmacokinetic parameters were: clearance, 0.27 x WT (L/h) on Day 21 and 0.233 x WT (L/h) at 6 months (ISV, 51%; interoccasion variability, 47%); central distribution volume, 47.6 L (ISV, 31%); intercompartmental clearance, 33.1 L/h (ISV, 128%); peripheral distribution volume, 724 L; duration of infusion, 0.64 (hours), and binding capacity, 0.0012 L/mg. The multiple regression analysis between free and total MPA concentration led to the following model: free MPA concentration (mg/L) = 0.31 + 0.02 total MPA concentration (mg/L) - 0.007 albumin (g/L). The large inter- and intravariability of MPA raises questions about the value of the use of therapeutic monitoring and limited sampling strategies as currently practiced. Moreover, none of the data presented here could justify measurement of free concentration for therapeutic drug monitoring.