The purpose of this study was to define the pattern of postnatal plasma cortisol concentrations during the first week of life in premature infants, and to evaluate the effect of developmental and clinical factors on this pattern. We measured plasma cortisol concentrations in the morning and afternoon on d 2, 4, and 6 in 120 premature infants (gestational age 24-36 wk) and examined the effects of gestational age, postnatal age, and illness. We described an inverse relationship between gestational age and cortisol concentrations, with the youngest infants having the highest random cortisol values (F = 5.14, p = 0.0073). Illness had a significant negative effect such that the cofactors ventilatory support pattern (F = 6.62, p = 0.0016) or "use of surfactant" (F = 6.63, p = 0.001) defined a pattern where cortisol values were lower in infants that had the highest ventilatory requirements or that received surfactant compared with values from those infants who did not have these requirements. The postnatal pattern in cortisol values depended on gestational age. Ill infants more than 27 wk gestational age increased their cortisol values from d 2 to d 6 although cortisol values decreased in well infants. These patterns resulted in a nonsignificant change over time for these age groups. In contrast, cortisol values significantly decreased from d 2 to d 6 in both well and ill infants that were less than or equal to 27 wk. We conclude that plasma cortisol concentrations in the premature infant are significantly correlated with gestational age and to markers of illness.(ABSTRACT TRUNCATED AT 250 WORDS)