Cerebral vasomotor responsiveness to 5% CO 2 or 100% 0 2 inhalation or hyperventilation was tested in groups of patients with either migraine or cluster headache. Regional cerebral blood flow was measured by the 133 Xe inhalation method before and during hypercapnia, hypocapnia or hyperoxia. A series of 79 headache patients were tested (18 classic, I complicated, 37 common, 16 cluster and 7 muscle contraction). For purposes of comparison, cerebral vasomotor responsiveness was likewise tested in 48 age and sex matched healthy volunteers. CO 2 responsiveness was expressed as D%Fg (gray matter blood flow) per mmHg PECO 2 (end-tidal CO 2 tension). Normal response was 3.6 ± 0.8% during 5% CO 2 inhalation and 3.0 ± 0.9% during hyperventilation. Inhalation of 100% O 2 inhalation reduced mean hemispheric Fg by 9.4 ± 5.4%. During the prodrome of migraine (N = 1) there was bilateral impairment of hemispheric CO 2 responsiveness. During the headache phase of migraine (N = 5) CO 2 responsiveness was impaired throughout both hemispheres but recovery occurred 24 hours after headache had subsided. During the headache-free interval of migraine (N = 21) there was excessive cerebrovascular CO 2 responsiveness (5.1 ± 2.2%). The cerebral hemisphere on the side of predominant head pain showed significantly greater excessive CO 2 response than the “non-headache” hemisphere. During cluster headache (N = 2) hemispheric CO 2 responsiveness was similarly impaired, but returned to normal immediately after the headache subsided. Unlike migraineurs, during cluster headache the cerebral vasoconstrictive response to 100% O 2 was excessive with marked reduction of Fg (33 ± 11%), with relief of headache. Two different types of abnormality in cerebral vascular receptor sites are hypothesized, which make it possible to differentiate migraine from cluster headache.