Context: Accurate hydrocortisone dosing in children with adrenal insufficiency is important to avoid the risks of over and under treatment including iatrogenic Cushing's syndrome and adrenal crisis. Objective: To establish a population pharmacokinetic model of hydrocortisone in children and use this to refine hydrocortisone replacement regimens. Design and methods: Pharmacokinetic study of hydrocortisone granules, available in 0.5, 1, 2 and 5 mg dose strengths, in 24 children with adrenal insufficiency aged 2 weeks to 6 years. Cortisol concentrations quantified by LC-MS/MS were used to refine an adult pharmacokinetic model to a paediatric population model which was then used to simulate seven different hydrocortisone treatment regimens. Results: Pre-dose cortisol levels were undetectable in 54% of the 24 children. The developed pharmacokinetic model had good predictive performance. Simulations for the seven treatment regimens using either three- or four-times daily dosing showed treatment regimens delivered an AUC(0-24h) within the 90% reference range for healthy children except in neonates where two regimens had an AUC below the 5th percentile. Cortisol concentrations at individual time points in the 24 h were outside the 90% reference range for healthy individuals in 50%, 55-65% and 70-75% for children, infants and neonates, respectively, with low cortisol levels being most prevalent. Conclusions: Current paediatric hydrocortisone treatment regimens based on either three- or four-times daily administration replicate cortisol exposure based on AUC(0-24h), but the majority of cortisol levels are above or below physiological cortisol levels with low levels very common before the next dose.