We used structural equation modeling (SEM) to test the hypothesis that childhood instrumental and vicarious learning experiences influence frequency of panic attacks in young adulthood both directly, and indirectly through their effects on anxiety sensitivity (AS). A total of 478 university students participated in a retrospective assessment of their childhood learning experiences for arousal-reactive sensations (e.g., nausea, racing heart, shortness of breath, dizziness) and arousal-non-reactive sensations (i.e., colds, aches and pains, and rashes). SEM revealed that learning history for arousal-reactive somatic symptoms directly influenced both AS levels and panic frequency; AS directly influenced panic frequency; and learning history for arousal-non-reactive symptoms directly influenced AS but did not directly influence panic frequency. These results are consistent with the findings of previous retrospective studies on the learning history origins of AS and panic attacks, and provide the first empirical evidence of a partial mediation effect of AS in explaining the relation between childhood learning experiences and panic attacks in young adulthood. Implications for understanding the etiology of panic disorder are discussed.