The heterogeneity of the clinical presentation of panic disorder (PD) has prompted researchers to describe different subtypes of PD, on the basis of the observed predominant symptoms constellation. Starting from a dimensional approach to panic disorder, an instrument to assess lifetime panic-agoraphobic spectrum (PAS) available in interview or self-report form (SCI-PAS, PAS-SR) was developed which proved to have sound psychometric properties and the ability to predict delayed response to treatment in patients with mood disorders. However, the structure of the instrument was defined a priori and an examination of its empirical structure is still lacking. Aim of the present report is to analyse the factor structure of the PAS taking advantage of a large database of subjects with panic disorders (N=630) assessed in the framework of different studies. Using a classical exploratory factor analysis based on a tetrachoric correlation matrix and oblique rotation, 10 factors were extracted, accounting overall for 66.3% of the variance of the questionnaire: panic symptoms, agoraphobia, claustrophobia, separation anxiety, fear of losing control, drug sensitivity and phobia, medical reassurance, rescue object, loss sensitivity, reassurance from family members. The first two factors comprise the DSM-IV criteria for panic disorder and agoraphobia. The other factors had received limited empirical support to date. We submit that these symptoms profiles might be clinically relevant for tailoring drug treatments or psychotherapeutic approaches to specific needs. Future perspectives might include the use of these factors to select homogeneous subgroups of patients for brain-imaging studies and to contribute to elucidating the causes and pathophysiology of panic disorder at molecular level.