Hostile-media-effect research has been built on the assumption that partisans are motivated to adjust their evaluations to accommodate preexisting attitudes that are especially sensitive to content. More recent literature, however, has suggested a rival hypothesis - that perceptions of media bias spring from social identity, keyed to source differences. To test these competing explanations, we designed an experiment around the teaching-alternatives-to-evolution controversy. Partisans from two adversarial groups read an op-ed article with manipulations of content (agreeable, balanced, or disagreeable) and source (in-group, neutral, or out-group). Both content and source influenced hostile media perceptions, but nuanced differences speak to the dynamic ways in which content, source, as well as group-level characteristics impact the processing and evaluation of news content.