Citizens across Latin America have reacted and mobilized collectively in diverse ways to resist and confront criminal violence and impunity, including taking justice in their own hands. The use of lynching stands out as a phenomenon through which victims and their communities attempt to capture and punish alleged criminals, who subsequently become victims of collective violence. Can lynching deter crime, or do they exacerbate it by further breaking down the rule of law? Large proportions of citizens across Latin America support lynchings and believe that they are effective in reducing crime, but to date there has been little research using causal inference methods to identify the effects of lynchings on crime at the local level. In this article, we attempt to fill that gap by analyzing a new time series dataset that combines novel data on lynching events coded from 47 Mexican news sources with existing survey and administrative data on crime, crime perceptions, and views of state authorities. We find that lynching events in Mexico are associated with short-term increases in property crime both in administrative records and in survey reports of crime victimization. These effects are concentrated in places where pre-lynching survey data shows low levels of community cohesion. These results are in line with a model in which lynchings are ineffective deterrents of crime and reduce the effectiveness of police responses to crime but are misperceived by citizens.