OBJECTIVE: Hyperventilation has been posed as an important symptom- producing mechanism in panic attacks. Some arguments and experimental findings, such as the possibility of inducing panic symptoms by voluntary hyperventilation in panic disorder patients, seem to favor this suggestion. This study was undertaken to clarify the role of hyperventilation in panic disorder. Long-term ambulatory measurement of transcutaneous arterial CO2 pressure (PCO2) offers an opportunity to test directly the co-occurrence of panic and hyperventilation under natural conditions. METHOD: Transcutaneous PCO2 was measured during three to four sessions of approximately 7 hours each in 28 panic disorder patients. Patients were instructed to expose themselves to fear-provoking situations and to press a button as soon as they experienced panic. One-half of the patients experienced one or more panic attacks during these sessions. RESULTS: A decrease in PCO2 was observed during only one of the 24 registered panic attacks that lasted at least 3 minutes. Even during this particular attack, the degree of hyperventilation was not impressive. CONCLUSIONS: These findings indicate that the hypothesis that hyperventilation is an important symptom-producing mechanism in panic may be dismissed.