IntroductionAntenatal administration of glucocorticoids to mothers at risk of preterm labour is known to improve production of surfactant in the neonate born prematurely, thus aiding maturation of lungs. Plasma cortisol concentration is expected to increase during significant stress [1,2] and recent studies in neonates by serial blood sampling have shown that cortisol is secreted in short pulses every 80 minutes [3]. These rapid pulsations are called ultradian rhythms and are caused by hypothalamic pituitary signalling. Arnold et al. [4] have shown that these secretory pulses in neonates exposed to antenatal glucocorticoids are shorter compared to neonates who have not been exposed to antenatal glucocorticoids.Plasma cortisol concentrations are affected by intrinsic factors (for example, immaturity of the hypothalamic-pituitary-adrenal (HPA) axis, physiological decline of plasma cortisol during the immediate post-partum period, effect of antenatal steroids on HPA axis) and extrinsic factors (for example, vaginal delivery, poor condition at birth, respiratory distress, mechanical ventilation, hypoglycaemia or infection) [5]. Extremely preterm infants may develop adrenal insufficiency in the early neonatal period requiring cortisol supplementation [6,7]. In these infants, hypotension may occur which is responsive to the administration of hydrocortisone [8]. Assessment of adrenal function in extremely premature infants by the administration of ACTH for a synacthen test is difficult because sequential blood sampling is required for determination of cortisol levels.Measurement of free cortisol is the best indicator of adrenal glucocorticoid secretion [9] because cortisol is more than 80% bound to cortisol binding globulins (CBG) in the circulation and therefore measurements of plasma cortisol can be compromised by conditions that alter CBG levels [10,11]. However, obtaining blood samples by venepuncture is a painful procedure, which might of itself cause high cortisol values [12] and free cortisol may be better tested by taking saliva samples [13] because salivary cortisol levels are known to accurately reflect the free fraction of cortisol in the body. Obtaining saliva is painless and non-invasive and therefore especially useful in neonates; it is unlikely to cause stress and thus salivary cortisol values are more likely to reflect the true plasma level [14].There is good evidence from studies in older children [15] and adults that salivary cortisol concentrations correlate well with plasma cortisol concentrations at baseline/rest as well as during exercise and during dynamic function tests involving the hypothalamic-pituitary-adrenal axis [16]. These studies suggest that plasma cortisol measurement can be replaced by salivary cortisol measurement. However, few studies [17] have been reported on salivary cortisol determination in neonates and particularly, extremely premature infants.We previously reported a novel method of collecting saliva, which is effective in obtaining a good quantity of saliva to enable relia...