Control over the offset of repeated administrations of 20% carbon-dioxide-e
nriched air was assessed in nonclinical participants (n = 30) reporting ele
vated levels of anxiety sensitivity-a population at an increased risk for e
xperiencing panic attacks and possibly developing panic disorder. In Phase
I, participants were randomly assigned to 1 of 2 conditions: one that permi
tted offset control over gas inhalation and one that did not. These conditi
ons were reversed in Phase II. Across phases, a lack of offset control resu
lted in greater self-reported anxiety compared with having control, althoug
h no significant differences were observed for heart rate. Whereas all part
icipants demonstrated a Stroop interference effect for general (e.g., coffi
n) compared with specific (e.g., dizzy) physical threat word types prior to
the first experimental phase, this effect persisted only for participants
who had offset control in Phase I. We discuss these results in relation to
the differential effects of offset control, with implications for better un
derstanding anxious responding during elevated bodily arousal.