Background: Conditioning bifunctional agent, busulfan, is commonly used on children before hematopoietic stem cell transplantation. Currently, at the Ramathibodi hospital, Bangkok, Thailand, initial dosing is calculated according to age and body surface area, and 7 samples per day are used for therapeutic drug monitoring (TDM). This study aimed to identify the best strategies for individual dosages a priori from patient characteristics and a posteriori based on TDM. Methods: The pharmacokinetic data set consisted of 2018 plasma concentrations measured in 135 Thai (n = 135) pediatric patients (median age = 8 years) and were analyzed using a population approach. Results: Body weight, presence of malignant disease, and genetic polymorphism of Glutathione S-transferase Alpha-1 ( GSTA1 ) were predictors of clearance. The optimum sampling times for TDM concentration measurements were 0.25, 2, and 5 hours after a 3-hour infusion. This was sufficient to obtain a Bayesian estimate of clearance a posteriori. Simulations showed the poor performance of a priori formula-based dose calculations with 90% of patients demonstrating a 69%–151% exposure interval around the target. This interval shrank to 85%–124% if TDM was carried out only at day 1 and to 90%–116% with TDM at days 1 and 3. Conclusions: This comprehensive study reinforces the interest of TDM in managing interindividual variability in busulfan exposure. Therapeutic drug monitoring can reliably be implemented from 3 samples using the Bayesian approach, preferably over 2 days. If using the latter is not possible, the formulas developed herein could present an alternative in Thai patients.