Tacrolimus (Tac) is the cornerstone of immunosuppression in pediatric liver transplantation (PLT). However, its use is complicated by side effects and lack of efficacy partly due to high pharmacokinetic (PK) variability. This thesis aims to better individualize Tac therapy in PLT according to factors accounting for this variability. We conducted three retrospective PK studies to identify and model the effect of these factors on Tac PK by using non-linear mixed effects modelling. We developed 3 population PK models. The two first studies focusing either on the one year or fifteen days post PLT showed that apparent clearance (CL/F) increases with graft weight/recipient weight ratio and time post PLT. In the last model, developed with bicentric data, Tac CL/F showed a 30% increase in case of donor CYP3A5 expression and a 29% decrease in case of donor CYP3A4*22 mutation. We also confrimed the inhibiting effect of fluconazole on CL/F according to the CYP3A5 status and proposed an algorithm for tacrolimus dose initiation.