Safety-relevant parameters are hypothesized to be important to the maintenance of pathological anxiety. The authors examined the effects of safety information and safety cues on anxious responding to a repeated 35% CO2 challenge in 31 patients with panic disorder. Patients were randomly assigned to one of three conditions: (a) safety information, (b) safety information plus a safety cue, or (c) no safety information. In the safety information group, patients received accurate information regarding the benign effects of the CO2 challenge. In the combined group, patients also received a safety cue that is salient for many patients with panic disorder (i.e., access to an anxiolytic pill during the challenge). The experimental manipulations did not differentially affect anxious responding following an initial challenge. However, after access to the anxiolytic was removed and the challenge procedure was repeated, those in the safety information alone condition showed lower subjective anxiety compared to those in the combined safety information/safety cue group. Findings suggest that safety information facilitates extinction of anxiety but only in the absence of safety cues.