The objective of this paper is to measure the effect of children’s television viewing on their performance in school. To this end, we use Japanese time use data to estimate ordered probit models with and without an instrumental variable. The instrumental variable is the number of viewable commercial broadcast channels in the prefecture where a child and his/her family live. We find that, all else being equal, the more television a child views, the more likely it is for the child’s mother to report negative performance of her child in school. However, our instrumental variable estimates provide no evidence that television viewing by children affects either the mother’s report of the performance of her child or the child’s self-report on the speed of lessons in school. This study provides no evidence that television viewing has a negative impact on children’s performance in school, although we should not overlook the possibility of weak instruments.