Immigration enforcement is increasingly integrated with local policing. This trend accelerated four years ago when the federal government launched “Secure Communities,” a program designed to check the immigration status of every person arrested by local police. The government views the program as an innovation that enhances the efficacy of crime control and immigration enforcement, while civil rights groups have decried it as an invitation to racial profiling by local police. This paper, part of a larger project providing the first large-scale empirical evaluation of Secure Communities, uses the pattern of the program’s activation to explore a central feature of criminal and administrative law that rarely lends itself to empirical examination — the role of discretion in policing. Constrained by limited resources, the agency staggered the program’s activation across counties, revealing the federal government’s priorities for implementation in the face of competing political and programmatic incentives. The data undercut the government’s claims that the program is all about making communities more secure from crime. Moreover, the fact that early activation in the program correlates strongly with whether a county has a large Hispanic population raises important questions about demographic profiling in immigration enforcement.