IT IS RELATIVELY seldom that the physician has an opportunity to witness an attack of cluster headache (Horton's syndrome). Consequently, the diagnosis generally is based on anamnestic data supplied by the patient. A definite method for provoking an attack would therefore be of value, from both the diagnostic-therapeutic and the pathophysiological aspects, to establish the mechanism of the actual headache. Horton1-3showed that subcutaneous injection of histamine can elicit an attack, and reported positive results in 69% of his cases. The corresponding incidence in Peters' series was 75%.4Several authors have, however, doubted the value of the histamine test. Thus, Gardner et al5found that a typical attack could be precipitated only in the occasional case, and Friedman and Mikropoulos6were not convinced of the specificity or value of the test. Although nitroglycerin has been mentioned as a provocative agent in connection with Horton's syndrome,