Objective: Based on adult data, a peak cortisol response R500 nmol/l to insulin-induced hypoglycaemia constitutes a normal. Age-specific reference ranges for basal morning cortisol have been developed for clinical use in the paediatric population. Such reference ranges are not clearly established for peak cortisol responses to insulin-induced hypoglycaemia despite limited data suggesting an effect of age on peak cortisol. The aims of this study were to assess factors affecting the cortisol response to insulin-induced hypoglycaemia in children and to determine whether the peak cortisol response was related to age. Design: The present study was a retrospective cohort study. Methods: Retrospective analysis of children and adolescents aged %18 years undergoing the insulin tolerance test with adequate hypoglycaemia was undertaken. Patients with hypopituitarism or severe hypothalamic-pituitary-adrenal axis impairment (peak cortisol value !400 nmol/l) or using systemic glucocorticoids were excluded. Results: Two hundred and twenty-three tests were analysed. Peak cortisol responses R500 nmol/l occurred in 183 (82%) tests. Age was negatively associated with peak cortisol responses (rZK0.15, PZ0.03). A peak cortisol response !500 nmol/l was significantly less common in patients aged !12 years (9/97 (9%) vs 31/126 (25%); PZ0.004). In children aged !12 years, the median (5th-95th centiles) peak cortisol values were 610 (480-806) nmol/l compared with 574 (442-789) nmol/l in children aged R12 years (P!0.004). Similarly, median cortisol increment was significantly higher in younger patients (301 nmol/l compared with 226 nmol/l (PZ0.0004)). Conclusions: Use of a single peak cortisol threshold in children of all ages is not appropriate and will result in overdiagnosis of adrenal insufficiency in adolescents.