Phenytoin has been decreasingly used because of the high interindividual variability in drug concentration and the narrow therapeutic window. Despite such drawbacks, phenytoin is still essential as a second-line therapy for status epilepticus when patients are resistant to benzodiazepines. This study aimed to develop a population pharmacokinetic model of phenytoin and to propose the optimal dose regimen of phenytoin in Korean epilepsy patients. Concentrations collected from electronic medical records for 117 patients, with 1 or 2 measurements per patient, were analyzed using NONMEM 7.3.0. One-compartment model with first-order elimination where allometry scaling was considered best described the data, yielding the estimates of V and CL of 68.19 (L) and 0.63 (L/h), respectively, for patients with a body weight of 60 kg. Covariate analyses showed that, after birth, clearance increases with age, reaching adult level at 4 years, and after 20 years, it decreases with age. Simulation results showed that the dosing interval should be reduced to achieve optimal dosing in neonates and infants, and the optimal dose required increases with weight. This work demonstrates that a model-based approach can serve as a useful tool to individualize phenytoin therapy.