Following George Floyd's killing in May 2020, Black Lives Matter (BLM) protesters called for police reform in the largest and most diverse protests in U.S. history. The police frequently responded to BLM protesters with force, raising the question of how protest policing may impact public attitudes toward the police. We examined this issue using data from a survey of voters in Portland, Oregon, which was conducted before the November 2020 election, when police reform was on the local ballot. We tested an extended version of the process-based model, contextualized in light of the dialogic model of legitimacy and scholarship on collective action frames. We found that 1) the perception that police have responded inappropriately to protesters is associated with voters’ support for police reform, even controlling for attitudes toward BLM, and 2) this association is indirect through the resonance of pro- versus anti-reform frames. Findings suggest that inappropriate protest policing may be the straw that breaks the camel's back, increasing the resonance of pro-reform frames, reducing the resonance of anti-reform frames, and motivating voters to vote for reform.