We report findings from an experiment where participants read a story about a speech that sharply criticized U.S. foreign policy. The story varied how elites framed the speech, the speaker’s apparent ethnicity, and the content of the speech. We assess how each of these factors affected not only tolerance judgments but also inferences about the speaker’s motives and the likely consequences of the speech—considerations that play a central role in free speech jurisprudence. Troublingly, we find that the effects of elite framing and the speaker’s apparent ethnicity are often comparable with the effect of the speech explicitly calling for violence. Our design also allows us to assess the extent to which the effectiveness of elite framing is constrained by "facts on the ground." However, we find little evidence that the framing effects we identify depend on the content of the speech or the speaker’s ethnicity.