We study the link between crime and extreme traffic congestion to estimate the psychological costs of traffic. Our empirical analysis combines police incident reports with observations of local traffic data in Los Angeles from 2011 to 2015. This rich dataset allows us to link traffic with criminal activity at a fine spatial and temporal dimension. Our identification relies on deviations from normal traffic to isolate the impact of abnormally high traffic on crime. We find that extreme traffic increases the incidence of domestic violence, a crime shown to be affected by emotional cues, but not other crimes. The result is robust to a variety of specifications and falsification tests. The results represent a lower bound of the psychological costs of traffic congestion, since most drivers stuck in traffic do not commit domestic violence but still bear some emotional costs.