A population-based study of active epilepsy in mentally retarded children identified 98 children, 6-13 years old. A biopathological origin was established in 66% of mildly and 92% of severely retarded children: a prenatal etiology was considered in 51% and 57%, a perinatal in 9% and 19%, a postnatal in 6% and 16% and an untraceable etiology in 34% and 8%, respectively. Severe mental retardation was more frequent in the peri- and postnatal groups (80% and 83%) than in the prenatal and untraceable groups (67% and 29%). Thirty-four pre- and perinatal optimal items were defined. Children with a prenatal etiology did not differ from controls in any of the periods. Children with a perinatal etiology had, compared with controls, higher proportions of non-optimal items successively increasing through the pre- and perinatal periods showing the accumulation of negative events.