Riots are extreme events, and much of the early research on rioting suggested that the decision making of rioters was far from rational and could only be understood from the perspective of a collective mind. In the current study, we derive and test a set of expectations regarding rioter spatial decision making developed from theories originally intended to explain patterns of urban crime when law and order prevail—crime pattern and social disorganization theory—and consider theories of collective behavior and contagion. To do this, we use data for all riot‐related incidents that occurred in London in August 2011 that were detected by the police. Unlike most studies of victimization, we use a random utility model to examine simultaneously how the features of the destinations selected by rioters, the origins of their journeys, and the characteristics of the offenders influence offender spatial decision making. The results demonstrate that rioter target choices were far from random and provide support for all three types of theory, but for crime pattern theory in particular. For example, rioters were more likely to engage in the disorder close to their home location and to select areas that contained routine activity nodes and transport hubs, and they were less likely to cross the Thames River. In terms of contagion, rioters were found to be more likely to target areas that had experienced rioting in the previous 24 hours. From a policy perspective, the findings provide insight into the types of areas that may be most vulnerable during riots and why this is the case, and when particular areas are likely to be at an elevated risk of this type of disorder.