This paper studies how legislators and their constituents respond to political violence using data from Twitter and roll-call votes, and employing both event study and difference-in- differences methods. Tweets from incumbent party legislators and tweets with a “hard-line” language receive higher engagement following rebel attacks. The incumbent party receives higher support in the legislature after attacks, but only when it has a hard-line military policy. In addition, politicians are more responsive to attacks which occur in their electoral district. I identify a set of potentially affected congressional votes, suggesting that these conflict- induced swings in incumbent support can have persistent policy consequences.