Background. There is increasing evidence that subthreshold forms of psychopathology are both common and clinically relevant. To enable classification of these subthreshold forms of psychopathology, it may be useful to distinguish not only a threshold for illness but also for health. Our aim was to investigate this with regard to panic.Method. Data were derived from the Netherlands Mental Health Survey and Incidence Study (NEMESIS), which is based on a large representative sample of the adult general population (18–65 years) of The Netherlands (n=7076). The Composite International Diagnostic Interview was used as a diagnostic instrument. By defining two thresholds, three groups were formed: panic disorder, subthreshold panic disorder and no-panic. These groups were compared using multinomial regression analysis, χ² and analysis of variance.Results. The 12-month prevalence of panic disorder was 2·2% while that of subthreshold panic disorder was 1·9%. Symptom profiles and risk indicators associated with panic disorder and subthreshold panic disorder were similar, and half of the risk indicators were more strongly associated with panic disorder than with subthreshold panic disorder. Subthreshold panic disorder occupied an intermediate position between panic disorder and no-panic with regard to the number of symptoms, the percentage of subjects with co-morbidity, and functioning.Conclusions. Subthreshold panic disorder is common, and seems clinically relevant, but is milder than panic disorder. These results thus support the use of a double threshold in panic. Further research should focus on the positioning of the thresholds, the course of subthreshold panic disorder and its treatment options.