I examine the link between IQ and an individual¡¯s decision to join a gang. Data from the NLSY97 and Project on Human Development in Chicago Neighborhoods (PHDCN) are used to estimate time-to-first gang participation. Results from a variety of models which account for sibling effects, neighborhood effects, and non-cognitive traits indicate low IQ is a robust predictor of gang participation. However, the PHDCN results reveal gang participation is affected by a person¡¯s relative IQ, with respect to one¡¯s neighborhood peers. Because the majority of trade and industry is underground, official statistics overlook that neighborhoods where gang activity is prevalent are often at full employment. If gangs provide security and enforce contracts where civil government does not, then low-IQ individuals may have comparative advantage in gang activities. Because gangs are often well-defined social groups within neighborhoods, cognitive traits could be expressed at the neighborhood level through this same economic channel.