The dosage regimen of cyclosporine (CsA) can be individualized in patients by means of a test dose (TD) method in the few days before bone marrow graft. To simplify the test dosing protocol, we used a Bayesian estimation (BE) of CsA clearance requiring population data and partial kinetic information for a given patient. In the first part, 42 patients, aged 11–47 years, were given a 2-h infusion of CsA (4 mg · kg-1). CsA concentrations were determined in several blood samples. The obtained concentration-time curves were fitted according to a three-compartment model. Maximum likelihood estimation (MLE) allowed CsA determination of pharmacokinetic parameters. Early population parameters of CsA were determined using 22 TD by a two-stage method. In the remaining 20 patients, individual pharmacokinetic parameters were also computed by BE procedure, taking into account the above determined population parameters and CsA concentration determinations from three blood samples drawn at 5 and 30 min and 3 h after the end of TD infusion. In the second part, 16 patients were used to evaluate performance of BE in directly predicting target concentration values. Finally, early population characteristics were updated on the basis of 42 patients. Statistical comparisons between MLE and BE estimates of CsA clearance and between concentrations predicted after BE and those experimentally obtained showed that three blood CsA determinations allowed accurate clearance estimation and target concentration predictions. The method presented here can be used to calculate an individual CsA dosage regimen in real time, thus improving patient comfort. Furthermore, this real-time dosage regimen adjustment is more suitable, as it makes it possible to avoid the dependence phenomena of CsA kinetics over a long time period.