BACKGROUND The National Comorbidity Survey is a nationally representative survey of the prevalences and correlates of DSM-III-R disorders in the US household population. METHODS Retrospective age-at-onset reports were used to study predictive relationships between lifetime panic and depression. RESULTS Strong associations were found between the lifetime prevalences of panic and major depressive episodes (odds ratios: for panic attacks with depression, 6.2; for panic disorder with depression, 6.8). These associations were not significantly influenced by the inclusion or exclusion of respondents with mania. Temporally primary depression predicted a first onset of subsequent panic attacks but not of panic disorder. Temporally primary panic attacks, with or without panic disorder and whether or not the panic was persistent, predicted a first onset of subsequent major depression. The associations between panic attack and depression were attenuated in models that controlled for prior traumatic life experiences and histories of other DSM-III-R disorders. CONCLUSIONS Lifetime panic-depression comorbidity characterizes most community respondents with panic disorder and a substantial few of those with major depression. The absence of a dose-response relationship suggests that primary panic attack is a marker, rather than a causal risk factor, of subsequent depression. Primary depression, in comparison, appears to be a genuine risk factor for secondary panic attacks. That primary depression predicts panic attacks but not panic disorder suggests that secondary panic is a severity marker of depression rather than a comorbid condition. These results are far from definitive because they are based on retrospective reports, lay-administered diagnostic interviews, and only 1 survey. However, they raise important questions that could lead to a fundamental rethinking of panic-depression comorbidity if they are replicated in future epidemiological and clinical studies.