The study was designed as a panel survey, covering a 3-year period (1970-73). Data were obtained from approximately 2,400 elementary school children (second through sixth grades) who were surveyed up to six times and from 800 teenage boys, who were surveyed up to five times. Interviews and questionnaires were administered to the respondents and samples of their parents and school teachers. The study measures purposive aggression, physical or verbal acts intended or known in advance to cause injury to others. A peer-nomination measure was used to determine aggression in the elementary school sample, and a self-report measure was used for the teenage sample. The measure of television exposure consisted of questions keyed to an extensive program checklist developed on the basis of pretests in which the respondent was asked to indicate the frequency of viewing of a large sample of specific programs. The findings show no significant association between violent television exposure and subsequent change in aggression. This differs from the findings of the experimental literature, which has shown that exposure to specific filmed segments influenced several types of aggression. These findings are not necessarily contradictory, since the experimental literature is concerned with short-term arousal and modeling effects. Long-term socialization effects in a real-life context, which the current study addresses, are not studied in the literature. 9 notes and 42 references