Self-monitoring of panic attacks and retrospective estimates of panic: discordant findings.

An event sampling method was used to study the frequency of panic attacks during treatment of agoraphobics. Results revealed a much lower incidence of panic attacks in agoraphobics according to self-monitoring than was expected on account of their retrospective estimation. When more stringent criteria for panic attacks are applied, retrospective overestimation becomes even more apparent. The implication of this finding for the classification of panic disorder patients is discussed.