The cognitive model of panic (Clark, 1988) suggests that panic attacks result from the catastrophic misinterpretation of bodily sensations rather than the sensations themselves. Anxiety sensitivity (AS) is fear of anxious bodily sensations (Reiss, 1991) and has implications in panic development, maintenance, and severity. Although previous work has demonstrated that AS amplifies symptoms in response to provocations, few have analyzed the role of AS in the relationship between panic symptoms and panic disorder severity. The purpose of this investigation was to determine if AS, a cognitive risk for panic, has an indirect effect on the association between self-reported panic symptoms and panic severity, both self-reported and clinicianassessed, among 67 treatment-seeking individuals with a primary diagnosis of panic disorder with or without agoraphobia. Data were analyzed using the bootstrapped conditional process indirect effects model. Results indicated that the overall total mediational effect on Panic Disorder Severity Scale (PDSS) was significant with evidence of partial mediation. The direct effect of Beck Anxiety Inventory (BAI) on PDSS remained significant although there was also a significant indirect effect of BAI via AS. Results showed a similar relationship when Clinician Severity Rating was the outcome. Moderation analyses were not significant. Therefore, AS was a significant partial mediator of the relationship between symptom intensity and panic severity, whether clinician-rated or self-reported. This investigation provides support for the importance of AS in panic, highlighting its importance but suggesting that it is not sufficient to explain panic disorder.