OBJECTIVE To derive a model describing carbamazepine (CBZ) clearance in children, in terms of individual patient characteristics. METHODS One hundred and eighteen steady-state serum carbamazepine concentration measurements were gathered during normal routine care of 72 compliant outpatients (2.3-16.3 years old). Levels were obtained from patients receiving monotherapy (55%), concomitant valproate (26%), or concomitant inducers (phenytoin, phenobarbitone; 19%). A one-compartment model was used to fit the data with the computer programme Nonlinear Mixed Effects Model (NONMEM). RESULTS Weight, age and concomitant medication were all important determinants of clearance. The final model for clearance (1.h-1) was: CL = [0.7(WT) 0.4] . M, where WT is patient weight (kg) and M is a scaling factor for concomitant medication, with a value of 1 for patients on CBZ monotherapy or concomitant valproate and 1.4 for those receiving concomitant inducers. For the purposes of this analysis, bioavailability (f) was assumed to be complete, i.e., f is thus included in the term CL. CONCLUSIONS CBZ clearance decreased with increasing age. As age and weight were correlated, either variable was a satisfactory predictor. The influence of both the inducers and valproate on CBZ clearance was as expected. This model, which describes clearance in terms of patient-specific details, can be used when predicting the maintenance dose required to achieve a target mean steady-state CBZ concentration in children.