Gangs have recently migrated from major metropolitan areas to suburban and rural communities throughout the United States (Maxson 1998; NAIGA 2005). This migration creates a need for further research and understanding of the gang phenomenon in the United States. One commonly studied aspect of gangs is their propensity to participate in homicide and violent behavior (e.g., Curry and Spergel 1988; Decker and Curry 2002). This tendency has been tested and written about in the literature; however, none of the research addresses the migration and dispersion of gangs throughout states. Therefore, new approaches are needed to better understand gangs and their behavior outside of urban areas. This study is the first state-level test of gang homicide variation. Previous gang homicide studies incorporated structural level variables; these, however, were limited to metropolitan and urban areas. In addition, the inclusion of structural level variables was commonly without an appropriate theoretical framework able to explain variation in gang homicide rates across states. Messner and Rosenfeld's (2007) institutional anomie theory provides this framework incorporating the interaction between the cultural ethos and social institutions of states. Gang homicides, as reported in the Uniform Crime Report Supplementary Homicide Reports from 2000, are used as the dependent variable in the multiple regression models. Cultural ethos is measured through economic decommodification, and education, polity, family, and religion are included as structural measures. Characteristics of a state's population are also included in the regression. These characteristics include: age structure, urban population, minority population, new immigrants, and incarcerated drug offenders. Additive and interactive relationships are tested using multiple regression. The results of the present study do not provide support for the theoretical model. Support was not found in either the interactive or additive models for any of the structural or cultural theoretical measures. Urban population and the state's age structure did provide empirical support in both the additive and interactive models. Race and immigrant status, as well as drug incarceration rates, did not have empirical support. This study concludes with an in-depth discussion of the unexpected findings, limitations, and future research ideas.