Adrenal insufficiency is suspected in some ill preterm infants. The aim of this prospective study was to compare serum cortisol concentrations during the first 2 wk of life of well preterm infants (group A) less than 30 wk of gestational age with the cortisol concentrations of ill preterm infants whose arterial hypotension-a potential sign of adrenal insufficiency-had been treated with catecholamine (group B), and the cortisol concentrations of ill preterm infants who had not been so treated (group C). Cortisol concentrations did not differ significantly between group A (240 nmol/l, 58-659; n = 46) (median, minimum-maximum) and group C (268 nmol/l, 58-1007; n = 25). Group B had a double-peaked distribution of cortisol. Two subgroups were formed by taking the highest cortisol level of group A as a threshold: group B1 (110 nmol/l, 41-378; n = 20) and group B2 (1200nmol/l, 764-1482; n = 8). The cortisol concentrations of group B1 were significantly lower (p = 0.00097) compared to the cortisol concentrations of the well preterm infants (group A). The severity of illness, which was quantified by two scoring systems, differed significantly among the groups (p < 0.003 for all comparisons) with the following sequence: A < C < B, but not between B1 and B2, as clinical variables were not different between the subgroups.