To establish using dried blood spot (DBS) as a surrogate to plasma for therapeutic drug monitoring (TDM) of carbamazepine (CBZ), we compared the population pharmacokinetic (PPK) estimates from concurrent DBS and plasma levels. The dose-concentration relationship, estimated parameter and variability were determined. A total of 98 observations from 97 people with epilepsy (PWE) were included in this study. Data was split into 3:1 ratio for the respective index group and validation group. Non-linear mixed effects regression with one compartment, first order absorption and elimination model was utilized. Covariates were screened for inclusion into final model via forward stepwise addition and backward elimination method. Predictive performances of the final models were assessed for bias and precision. The typical clearance for CBZ was estimated to be 5.85 and 5.68 L/h from plasma and DBS concentrations, respectively. The final models for clearance estimates obtained from plasma concentrations (Cplasma) included total daily CBZ dose per unit weight (DD) and sex while from DBS concentrations (Cdbs) included only DD. The final models were both precise and non-bias. The developed PPK models had comparable estimates, errors and predictive performances. Our findings suggest that Cplasma and Cdbs could be used interchangeably for pharmacokinetic studies of CBZ.