OBJECTIVE--To examine the characteristics of cold-induced headaches in a group of migraine patients, to compare these with their usual migraine headaches and with cold-induced headaches in a control population. DESIGN--Subjects completed a structured questionnaire recording previous headache history along with the characteristics of any headache produced during supervised palatal and pharyngeal application of ice cream. SUBJECTS--70 consecutive patients attending the City of London Migraine Clinic, and 50 pre-clinical medical and dental student volunteers from Queen Mary and Westfield College. RESULTS--27% of the migraine patients and 40% of the students reported previous ice cream headaches. 17% of the migraine patients and 46% of the students developed headache following palatal application or a swallow of ice cream. Typically the headache was of early onset (x = 12.5s) and short duration (x = 21s), with a tendency for anterior headache on the same side as a palatal stimulus, and bilateral headache following an ice cream swallow. However, a significant minority experienced a previously unreported headache of late onset (x = 102s) and long duration (x = 236s) which tended to occur particularly after swallowing ice cream and to be less well localised to the side of the cold stimulus. Ice cream appeared not to be a common trigger for migraine, and there was no significant correlation between site of ice cream headache and usual site of migraine. CONCLUSIONS--These findings confirm that cold stimulation of the palate or pharynx commonly produces a headache. In contrast to previous studies, our results suggest that the 'ice cream headache' is less common in migraine patients than the general population. A similar pattern of headache was produced in both migraine patients and controls, and apart from the few for whom an ice cream headache may trigger a migraine, the ice cream headache seems not to have any special significance for migraine patients.