There is evidence that excessive cortisol secretion in depressed patients might result, in part, from an enhanced adrenocortical sensitivity to corticotropin. This phenomenon has been examined using the cosyntropin (alpha1-24-corticotropin) stimulation test. Most studies have used supramaximal doses of cosyntropin administered in the morning, when adrenal sensitivity to corticotropin is at its maximum. This could partially obscure subtle differences in adrenocortical sensitivity in depression that might otherwise be evident at lower cosyntropin doses given later in the day. To test this hypothesis, we administered two consecutive cosyntropin tests on separate occasions employing a submaximal 0.05-microgram/kg dose and a maximal 0.2-microgram/kg dose. The cortisol centered cumulative response over 240 minutes was measured after each test in 12 depressed patients (7 melancholic, 5 nonmelancholic) and 6 healthy volunteers. When the difference in mean cortisol centered cumulative response values was determined, healthy controls demonstrated a significant increase in cortisol centered cumulative response, while the nonmelancholic patients had a less robust increase in cortisol centered cumulative response. In contrast, the melancholic patients demonstrated cortisol responses similar to those of the healthy subjects after each cosyntropin dose, suggesting an enhanced adrenocortical sensitivity to corticotropin. These data support the hypothesis that increased glucocorticoid secretion in depression may result from abnormalities at several sites within the hypothalamic-pituitary-adrenocortical axis.