We examine the causal effect of unionization on firm innovation. To establish causality, we use a regression discontinuity design relying on ‘locally’ exogenous variation generated by elections that pass or fail by a small margin of votes. Passing a union election leads to an 8.7% (12.5%) decline in patent quantity (quality) three years after the election. A reduction in R&D expenditures, reduced productivity of current and newly hired inventors, and departures of innovative inventors appear plausible mechanisms through which unionization impedes firm innovation. Our paper provides new insights into the real effects of unionization and has important implications for policy makers when they alter union regulations or labor laws to encourage innovation.