Acyclovir is a primary treatment for central nervous system (CNS) infections caused by herpes simplex virus (HSV) and varicella-zoster virus (VZV). However, patient outcomes remain suboptimal with high mortality and morbidity, following current dosing guidelines. Given the lack of alternative therapies, there is a pressing need to optimize acyclovir dosing, especially since initial regimens were developed in the 1980s with incomplete pharmacokinetic data in the CNS. This study aimed to evaluate both current and alternative acyclovir dosing regimens using a full Bayesian physiologically-based pharmacokinetic (PBPK) model tailored for viral encephalitis. We developed a CNS PBPK model to simulate acyclovir concentrations in plasma, brain extracellular fluid (ECF), and subarachnoid space (SAS). Drug efficacy was assessed using two pharmacokinetic targets, 50%fT>IC50and Cmin>IC50, with a safety threshold set at 25 mg/L in plasma. The standard dosing regimen (10 mg/kg TID) yielded sufficient acyclovir exposure in plasma, brain extracellular fluid (ECF), and subarachnoid space (SAS) compartments based on the 50%fT>IC50target. However, it did not consistently meet the Cmin>IC50target, indicating potential suboptimal exposure in these compartments when evaluated against this criterion. Notably, a higher probability of target attainment (PTA) was generally observed in the brain ECF and SAS compared to plasma. Increasing the dosing frequency to QID improved target attainment but exceeded the toxicity threshold at 20 mg/kg. Our findings suggest that a dosing regimen of 10 mg/kg or 15 mg/kg QID may offer a more effective and safer approach for managing CNS infections compared to the other tested alternative dosing regimens.