Recent reviews of studies concerning panic attacks in adolescents have emphasized research limitations, noting problems of validity, reliability, and lack of normative data. To address some of these limitations we evaluated two methods of panic ascertainment (questionnaire versus interview), reliability of interview-determined panic, and clinical correlates of panic symptoms in a large sample (N = 1013) of early adolescent girls. The 5.4% of the sample who, when interviewed, reported ever experiencing a panic attack scored significantly higher on measures of depression, anxiety sensitivity, and alcohol use, but were not more avoidant than others. Using the interview as the standard, the questionnaire had a specificity of 81% and a sensitivity of 72%. Adolescents do experience panic attacks-whether identified by questionnaire or interview-although for many the attacks may not be salient. Longitudinal studies are required to determine those qualities of nonclinical panic (severity, context, interpretation/attribution), which render some episodes as clinically meaningful.