We found moderate correlations between reported television viewing at different ages (table 1). These data indicate that viewing habits established in childhood might persist into early adulthood. This in itself is a cause for concern. However, it suggests that child and adolescent viewing might be associated with poor adult health because it is correlated with viewing in adulthood. Because we did not obtain information on television viewing at age 26 years, we could not test this hypothesis directly. However, controlling for television viewing at 21 years of age did not eliminate the association between earlier viewing and age-26 outcomes. Indeed, age-26 health was better predicted by television viewing in childhood and adolescence than at age 21 years. Although viewing time estimates at age 21 years might have been less accurate than the composite measure at age 5 to 15 years, this finding suggests that television viewing during childhood and adolescence is associated with long-lasting detrimental effects.