The self-report questionnaire is the most common data collection procedure used in normative headache investigations. However, reliability data for such an instrument is notably lacking in the headache literature. In the present study, the reliability of questions commonly used in headache questionnaire assessment was evaluated. One hundred and one college students were given a headache questionnaire twice, with three months separating the administrations. As a group, the subjects were very reliable as indicated by highly significant (p <.001) correlations. However, differences were noted when the subject population was divided into “problem headache” and “non-problem headache” groups on the basis of reported headache frequency and severity. There was a tendency for the “problem headache” group to be less reliable reporters of headache severity and accompanying symptoms. Results are discussed with regard to the need for more research into symptom reliability and its facilitative effect on the development of a standard headache assessment instrument.