The time relationship between DTP immunization and infantile spasms (IS) onset was examined using three models—association, temporal shift, and no-effect—and the case/control data from the National Childhood Encephalopathy Study (NCES). Infantile spasms cases classified as being previously abnormal (e.g. tuberous sclerosis complex patients) showed a no-effect relationship, whereas those classified as previously normal suggested a fit to the temporal shift model, i.e. no increase in number of cases but a shortening of time to onset of seizure. No data fit the association model. Analyses for vaccine complications should examine for temporal changes (i.e. temporal shift) in addition to increased risks.