Identify persons with epilepsy by first looking for prescriptions for particular antiseizure drugs. Follow these prescriptions from the pharmacies to the physicians who wrote them for patients. Ask the physicians whether the patients have epilepsy. Finally, contact the patients who do have epilepsy to elicit information about the impact of that condition on their lives. With these steps, it may be possible to carry out successfully a probability survey of epilepsy in the United States population. To learn more about this approach, a field test was funded by the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) of the Public Health Service. From 1978 through 1982, the work was planned, carried out, and evaluated by Research Triangle Institute, Research Triangle Park, NC. Epilepsy is a sensitive topic to ask about in a survey. Also, the condition is sufficiently rare to render ordinary survey approaches inefficient. Even if rarity were not an issue, there would be the problem of response error because a person with epilepsy does not, as a rule, have much clinical information on his or her condition. Better information lies with the physician who provides the care, but many physicians are busy with their practices. Furthermore, their record systems are usually not designed for easy retrieval of information, unless the names of patients are available. In the survey approach considered here, the names of patients are obtained through a random sampling of prescriptions for antiseizure drugs. The field test was divided into three phases with special activities reserved for each. The most important problem confronted was how to safeguard the confidentiality of relationships between pharmacist and patient and between physician and patient.Special guidelines on confidentiality were put into effect for the data collection. These guidelines,however, contributed to serious problems of nonresponse-especially for physicians. This article provides a brief account of the field test, including a rationale for the survey strategy of finding cases of epilepsy through prescriptions for antiseizuredrugs.