In 1988 the International Headache Society (IHS) introduced new diagno
stic criteria for headaches and craniofacial pain. Since headaches can
be diagnosed solely on the basis of information provided by the patie
nt, it is essential that the criteria are reproducible and consistent.
Two neurologists evaluated the clinical records of 100 consecutive ou
tpatients and transferred the data on headache and associated phenomen
a to a form designed to reflect the IHS criteria. Interobserver concor
dance (kappa statistics) in the application of the diagnostic criteria
of primary headaches was: (i) ''perfect'' to ''substantial'' for the
first IHS digit, being kappa = 1.0 for cluster headache and paroxysmal
hemicrania; kappa = 0.88 for migraine; kappa = 0.75 for tension-type
headache; (ii) ''almost perfect'' to ''substantial'' for the second di
git (kappa = 0.94 for cluster headache; kappa = 0.90 for migraine with
aura; kappa = 0.81 for episodic tension-type headache; kappa = 0.78 f
or migraine without aura; kappa = 0.71 for chronic tension-type headac
he; kappa = 0.66 for cluster headache-like disorder not fulfilling the
criteria; (iii) ''moderate'' for migrainous disorder (kappa = 0.48) a
nd headache of the tension-type (kappa = 0.43) not fulfilling the crit
eria. These results show that the IHS diagnostic criteria are satisfac
torily applicable to high quality medical records abstracted by experi
enced neurologists.