A modified cohort method has been proposed for estimating the relative incidence of rare adverse reactions after vaccination. The method requires only a sample of the cases, thus avoiding the need for following large population cohorts or selecting controls. This case series method has statistical power equivalent to that of the full cohort method when the risk periods after vaccination are short and vaccine coverage is high. The method also eliminates confounding by variables associated with both the outcome and avoidance of vaccination. In this paper, the cohort, case-control, and case series methods are reviewed, and their underlying assumptions and performances are compared. Theoretical results are illustrated using data on febrile convulsions after measles-mumps-rubella vaccination in the United Kingdom.