We examined the effects of hyperventilation and other manipulations of respiratory pace on parasympathetic nervous system function and subjective reactivity in 15 patients with panic disorder, 15 patients with social phobia, and 15 healthy control subjects. After a 30-minute rest period subjects completed a 2.5-minute trial of each of hypoventilation, normoventilation, and hyperventilation. Trials were separated by a 3 minute inter-trial interval. Incidence of panic attacks, symptom severity, vagal tone, heart rate, end-tidal carbon dioxide level, and respiratory frequency were measured throughout. Resting physicological measures did not differ between groups. Each respiratory manipulation resulted in the expected physiological changes (e.g., hyperventilation attenuated vagal tone), however, groups did not exhibit differential physiological reactivity to the manipulations. There were no panic attacks reported during either the hypoventilation or normoventilation phases; however, two social phobic subjects (13.3%) and two panic disorder patients (13.3%) reported panic attacks during hyperventilation. Although both groups of anxiety patients reported greater severity of hyperventilation-induced symptoms than did control subjects, symptom severity did not correlate significantly with vagal tone or heart rate. These results suggest that parasympathetic function is unlikely to be aberrant in PD patients and that diminished parasympathetic activity is not sufficient for the experience of panic attacks.