Context: Various cosyntropin doses are used to test adrenal function in premature infants, without consensus on appropriate dose or adequate response. Objective: The objective of this study was to test the cortisol response of extremely low birth weight infants to different cosyntropin doses and evaluate whether these doses differentiate between groups of infants with clinical conditions previously associated with differential response to cosyntropin. Design: The design was a prospective, nested study conducted within a randomized clinical trial of low-dose hydrocortisone from November 1, 2001, to April 30, 2003. Setting: The setting was nine newborn intensive care units. Patients: The patients included infants with 500-999 g birth weight. Intervention: The drug used was cosyntropin, at 1.0 or 0.1 μg/kg, given between 18 and 28 d of birth. Main Outcome Measure: We measured the cortisol response to cosyntropin. Results: Two hundred seventy-six infants were tested. Previous hydrocortisone treatment did not suppress basal or stimulated cortisol values. Cosyntropin, at 1.0 vs. 0.1 μg/kg, yielded higher cortisol values (P<0.001) and fewer negative responses (2 vs. 21%). The higher dose, but not the lower dose, showed different responses for girls vs. boys (P = 0.02), infants receiving enteral nutrition vs. not (P < 0.001), infants exposed to chorioamnionitis vs. not (P = 0.04), and those receiving mechanical ventilation vs. not (P = 0.02), as well as a positive correlation with fetal growth (P = 0.03). A response curve for the 1.0-μg/kg dose for infants receiving enteral nutrition (proxy for clinically well infants) showed a 10th percentile of 16.96 μg/dl. Infants with responses less than the 10th percentile had more bronchopulmonary dysplasia and longer length of stay. Conclusions: A cosyntropin dose of 0.1 μg/kg did not differentiate between groups of infants with clinical conditions that affect response. We recommend 1.0 μg/kg cosyntropin to test adrenal function in these infants.