Dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis may play a role in panic disorder. HPA studies in patients with panic disorder, however, have produced inconsistent results. Seeking to understand the inconsistencies, we reexamined endocrine data from four studies of patients with panic disorder, in light of animal data highlighting the salience of novelty, control, and social support to HPA axis activity. Patients with panic disorder were studied (1) at rest over a full circadian cycle, (2) before and after activation by a panicogenic respiratory stimulant (doxapram) that does not directly stimulate the HPA axis, and (3) before and after a cholecystokinin B (CCK-B) agonist that is panicogenic and does directly stimulate the HPA axis. Patients with panic disorder had elevated overnight cortisol levels, which correlated with sleep disruption. ACTH and cortisol levels were higher in a challenge paradigm (doxapram) than in a resting state study, and paradigm-related ACTH secretion was exaggerated in patients with panic disorder. Panic itself could be elicited without HPA axis activation. Patients with panic disorder showed an exaggerated ACTH response to pentagastrin stimulation, but this response was normalized by prior exposure to the experimental context or psychological preparation to reduce novelty and enhance sense of control. Novelty is one of a number of contextual cues known from animal work to activate the HPA axis. The HPA axis abnormalities seen in patients with panic disorder in the four experiments reviewed here might all be due to exaggerated HPA axis reactivity to novelty cues. Most of the published panic/HPA literature is consistent with the hypothesis that HPA axis dysregulation in panic is due to hypersensitivity to contextual cues. This hypothesis requires experimental testing. Depression and Anxiety 24:66–76, 2007. Published 2006 Wiley-Liss, Inc.