The theory that smoking could be implicated in the pathogenesis of migraine has been investigated in a prospective survey by the use of a questionnaire and the measurement of carboxyhaemoglobin (COHb) levels in patients attending a migraine clinic. There was a low incidence of smokers in patients attending for routine consultations and only a small percentage of patients thought that smoking was a cause of their headaches. This percentage was approximately the same both for migraine and for tension headaches. In patients attending with acute headaches, the incidence of smoking was also low, and raised COHb levels were found only in the smokers, most of whom did not consider smoking a cause of their symptoms. It is concluded that smoking is unlikely to be a factor in the aetiology or the exacerbation of migraine.