Background: Although panic disorder is generally considered to be a chronic condition, little is known about the natural history of panic attacks and the factors predicting the prognosis of panic. It is expected that data derived from the community show more variation in the prognosis of panic and that Subjects with less severe, 'subthreshold panic disorder' have a better prognosis. Methods: Usinga large, representative population-based study, the 2-year course of panic attacks and its predictors were investigated among 155 subjects with panic disorder and subthreshold panic disorder. Presence and frequency of attacks were studied using structured interviews. Putative risk-indicators of chronicity included socio-demographics, psychobiological, environmental, psychiatric and panic-related factors. Results: Thirty-nine percent of those with panic disorder and 17% of those with subthreshold panic disorder reported panic episodes during more than 75% of the observed time periods in the Life Chart Interview. Forty-three percent of those with panic disorder and 14% of those with subthreshold panic disorder reported over 24 attacks per 3-month period. Male gender, severity of panic and agoraphobia predicted a high proportion of time spent in panic episodes. Low self-esteem, limited positive life events and severity of panic predicted highly frequent attacks. Limitations: The sample size was small, thereby limiting statistical power. The reported Outcome may have overestimated chronicity. Conclusions: As was expected, the 2-year prognosis of panic was more varied than is often found in clinical studies. Although Subthreshold panic disorder had a more favourable prognosis than full-blown panic disorder, a sizeable proportion of those with subthreshold panic disorder had an till favourable prognosis.