The contributions of the Rochester (Minnesota) Epidemiology Project and Mayo Clinic studies in the evaluation of the causes of epilepsy are summarized. The development of the unique population-based medical records-linkage resource is chronicled, and the measures of occurrence and association used in epidemiologic study designs to assess the causes of epilepsy are presented. The historical cohort design is optimal for the study of the relationship between common central nervous system insults and seizures, and case-control studies are best used for analysis of relatively rare seizure disorders. The major findings about the etiologic roles of traumatic brain injuries, central nervous system infections, cerebrovascular disease, brain tumors, neurodegenerative diseases, developmental disabilities, perinatal insults, and familial factors are discussed. The role of genetic factors in epilepsy has been controversial, perhaps because of the numerous causes of seizures and their episodic nature. Both potential environmental and genetic causes will continue to be assessed.