Summary. Although numerous estimates of the prevalence of seizure disorders in populations in the less developed world have now been published, these estimates are difficult to interpret due to lack of comparability of study methods and criteria for case definition. The results reported in this paper are from a large, collaborative study of disabilities in 2- to 9-year-old children in which standard research procedures and case definitions were used in three diverse populations (located in Bangladesh, Jamaica and Pakistan). A two-phase study design (screening followed by professional evaluations) was used in this study allowing for the professional evaluation to serve as the criterion in the estimation of prevalence, even for rare disorders. As a result, the prevalence estimates reported here have a high degree of comparability across populations and exhibit unusually strong validity for population surveys. Febrile seizures were the most common type of seizure history in all three populations, with point estimates of lifetime prevalence ranging from 10.9 to 62.8 per 1000. The lifetime prevalence rates of epilepsy (recurrent unprovoked seizures) ranged from 5.8 to 15.5 per 1000. Lifetime prevalence rates of neonatal, all provoked and all unprovoked seizures, as well as estimates of the prevalence of active epilepsy, are also reported.