Numerous clinical trials have demonstrated the efficacy of cognitive-behavioral treatment (CBT) for panic disorder. However, studies investigating the mechanisms responsible for improvement with CBT are lacking. The authors used regression analyses outlined by R. M. Baron and D. A. Kenny (1986) to test whether a reduction in fear of fear (FOF) underlies improvement resulting from CBT. Pre- and posttreatment measures were collected from 90 CBT-treated patients and 40 wait-list control participants. Overall, treatment accounted for 31% of the variance in symptom reduction. The potency of FOF as a mediator varied as a function of symptom facet, as full mediation was observed for the change in global disability, whereas the effects of CBT on agoraphobia, anxiety, and panic frequency were partially accounted for by reductions in FOF. Clinical implications and future research directions are discussed.