Mycophenolate mofetil (MMF) is increasingly used for the treatment of autoimmune diseases (AID). In renal transplant recipients, it has been demonstrated that adjustment of the MMF dose according to the area under the plasma concentration versus time curve (AUC) of mycophenolic acid (MPA), the active moiety of MMF, improves clinical outcome. The aim of this study was to develop a maximum a posteriori Bayesian estimator (MAP-BE) to estimate MPA AUC(0-12) in patients with AID using a limited number of samples. The predictive performance of the MAP-BE was compared with a multiple linear regression method. Full MPA concentration versus time curves were available from 38 patients with AID treated with MMF. Nonlinear mixed-effect modeling was used to develop a population pharmacokinetic model. Patients were divided in an index and a validation data set. The pharmacokinetic model derived from the index data set was used to develop several MAP-BEs. The Bayesian estimators were used to predict AUC(0-12) in the validation data set on the basis of a limited number of blood samples. The bias and precision of these predictions were compared with those of limited sampling strategies developed with multiple linear regression. The absorption of MPA was described with 2 first-order processes with a short and a long lag time and a subsequent first-order elimination. The 2-compartment model accounted for the entero-hepatic recirculation of MPA as well. Using 1-4 samples, MPA AUC(0-12) was adequately estimated by the MAP-BE. Bias (-5.5%) was not significantly different from zero, and precision was below 27%. The predictive performance of the multiple linear regression method was comparable. In conclusion, MAP-BEs were developed for the estimation of MPA AUC(0-12) in patients with AID. The predictive performance was good and comparable to those of the multiple linear regression method. Due to its flexibility with respect to sample times, the MAP-BE may be preferred over the multiple linear regression method.