In this study, three waves of data from the Panel Study of Income Dynamics’s Child Development Supplement were used to examine patterns of children’s early TV exposure and its influence on middle childhood and adolescence. Analyses examined the pathways of influence depending on whether a dosage (hours of exposure) or diet (proportion of content to total TV time) variable was used. Results revealed that, in a dosage model, violent hours of early TV exposure were associated with decreases in independent reading and increases in externalizing behavior problems, but these did not predict later achievement. Early educational TV amount of exposure was unassociated with intermediate behaviors and was only related to later math achievement. In the diet model, total time spent viewing television was related to decreases in independent reading and prosocial behavior and increases in internalizing and externalizing behavior problems whereas violent TV diet was no longer related to any outcome. Educational TV diet, on the other hand, was positively associated with reading and math achievement. When the effects of early educational TV viewing were modeled on achievement over time, positive associations between educational TV diet and achievement in reading and math endure over time in significance and magnitude. These results suggest two conclusions: educational television viewing (particularly diet) is positively associated with achievement 10 years later even while controlling for socioeconomic status. Second, the methodological issue of which type of viewing variable to use is one that may have serious implications for the findings of research on media impact.