A self-administered questionnaire designed to identify people with transient cerebral ischemia (TIA) in large population groups was distributed to approximately 10,000 elderly persons living in public and private retirement facilities in the United States. The population resided in eight cities, and 74% completed the survey. Transient focal neurological symptoms compatible with a diagnosis of TIA within the previous year were reported by 6.4% of the respondents. Uncertain manifestations of TIA, i.e., dizziness, light-headedness, or loss of balance, were noted by an additional 15.4%; the remaining 78.2% had no manifestations of this disorder. A clinical history for TIA was taken by a neurologist in a sample of 1,712 respondents to determine the validity of the questionnaire responses. Thirty (7.1%) of 423 persons presumed by the questionnaire to have TIA met the standard clinical criteria for the diagnosis. An additional 8 cases of TIA were found by the neurologist among the 1,289 interviewed persons presumed by the questionnaire to have uncertain TIA or no evidence of the disorder. The twelve-month period prevalence of TIA was estimated to be 8.2 per 1,000 in the total respondent population and 5.8 per 1,000 among those without a history of prior completed stroke. Despite the high proportion of false-positive neurological symptoms of TIA elicited by the questionnaire, this survey instrument nevertheless screened out for further study the 6% of the population in whom 55% of the TIAs occurred.