This study examined the effect of having a safe person present on artificially induced anxiety following a biological challenge among panic-disordered patients. Anxiety symptoms were induced using a 5.5% CO2-inhalation procedure. Panic patients underwent the inhalation procedure either in the presence or absence of their safe person. Nonanxious controls underwent the procedure without a safe person. Panic patients exposed to CO2 without their safe person present reported greater distress, a greater number of catastrophic cognitions, and a greater level of physiological arousal than did panic patients exposed with their safe person. The latter group did not differ from controls on most measures at postexposure. The attenuation of self-reported anxiety and catastrophic cognitions is consistent with the safety-signal theory and the cognitive model of panic, respectively. The results, however, are inconsistent with a biological model of panic. © 1995 American Psychological Association.