The demographic, clinical and therapeutic features of the respiratory subtype of panic disorder (PD) versus the non-respiratory subtype were studied in a prospective design. Sixty-seven PD outpatients (DSM-IV), who had previously been categorized into respiratory (n=35) and non-respiratory (n=32) subgroups, were openly treated with clonazepam for a 3-year period. The principal measure of efficacy was the number of panic attacks, obtained from the Sheehan Panic and Anticipatory Anxiety Scale. In the first 8 weeks of treatment (acute phase), the respiratory subtype group had a significantly faster response to clonazepam. During the follow-up (weeks 12-156), the two subgroups did not differ significantly in the number of panic attacks experienced from baseline to end point. Patients in the respiratory subtype were characterized by a later onset of disorder and a family history of PD. Patients in the non-respiratory subgroup had a significantly higher number of past depressive episodes than those in the respiratory subgroup. The respiratory subgroup had a faster response after 8 weeks of treatment and an equivalent response in the 3-year follow-up period. Clonazepam had a sustained therapeutic effect over the entire treatment period.