Background: Fentanyl is widely used for sedation in preterm neonates requiring artificial ventilation. Neonates are especially vulnerable to side effects such as respiratory depression with oversedation which has a direct concentration-effect relation. The dosage regimen is usually derived from adults employing an extrapolation based on body weight which cannot accurately determine organ maturation in neonates. There is a necessity for model-based precision dosing of fentanyl that includes maturation. This study aimed to develop a pragmatic physiologically based pharmacokinetic (PBPK) model of fentanyl and to evaluate its prediction accuracy in neonates. Methods: PBPK modeling and simulation were performed using SimCYP Population-based ADME Simulator version 20. The fentanyl compound model was developed based on physicochemical and pharmacokinetic parameters reported in the literature. The model was validated using observed values from 25 clinical studies on adults and 4 on pediatrics. Serum samples of 14 neonates were collected as routine care. Serum fentanyl concentration was measured by liquid chromatography-mass spectrometry. Results: 100% of the predicted values of serum fentanyl concentration in adult and pediatric patients met the 2-fold acceptance criterion. Mean error (ME) and root-mean-square error (RMSE) for all predicted values were 0.28±0.15%, and 1.16%, respectively. In our neonatal patients, 95.5% of the predicted values were within the 2-fold acceptance criterion with ME and RMSE of 0.05±0.075%, and 0.34%, respectively. Median predicted fentanyl AUC value in patients with severe oxygen desaturation (sODS) (percutaneous oxygen saturation <80%) was higher than non- or moderate- ODS (nmODS) patients (ODS: 213 ng/mL·h, nmODS:25.2 ng/mL·h, P=0.043), while no significant difference was observed in fentanyl clearance depending on ODS (sODS: 1.33 L/h, nmODS: 1.39 L/h, P=0.35). Conclusion: This study shows that PBPK modeling can provide a precise prediction of serum fentanyl concentration in neonates. This model may be a useful tool for dose optimization and individualization strategies to avoid oversedation.