Clinical data yielded measurements of incidence and prevalence for seizure disorders on Guam that were underestimates of the values obtained by other methods. In this community, house-to-house survey by nonmedical or paramedical personnel followed by clinical appraisal of positive responders was necessary to yield more reliable estimates of incidence and prevalence. Using criteria adopted from Kurland's Rochester study, NINDS workers identified 38 active cases in four villages studied intensively, and 140 active cases from clinical records relating to the rest of Guam. In the survey villages, the age-adjusted incidence rate was 35 per 100,000 per year, of which one third were symptomatic cases; the age-adjusted prevalence rate was 542 per 100,000 for 1968. If clinical data alone were used, there was severe underreporting of idiopathic cases while ascertainment of symptomatic seizure disorders was nearly complete.