Abstract The effects of partisan media on political knowledge are theoretically ambiguous. Knowledge effects are important because of their close connection to welfare effects, but the existing empirical literature on knowledge is limited. We study the knowledge effects of the Fox News Channel. Following DellaVigna and Kaplan (2007), we exploit naturally random variation in Fox's availability to identify causal effects. We use knowledge survey data from 2000, 2004 and 2008; our final sample has nearly one million question-level observations. We first confirm and expand on previous findings of Fox effects on voting. We then present an array of results from our knowledge analysis. While average effects (across issues), over the full time-frame are near-zero and most precise, we find evidence of positive effects both for issues that were more favorable to Republicans and for issues that Fox covered more often, and negative effects for issues Fox neglected. We also present evidence of Fox being associated with a decline in newspaper readership.