Countering violent extremism (CVE) policies are an outgrowth of central government counter-terrorism efforts that aim to curb public support for terrorism and pre-emptively deter individuals from committing violence. The literature has heavily criticized CVE programs, yet central governments continue to expand them, tasking local governments and police with engaging predominately Muslim minority communities and establishing intervention efforts for individuals considered to be “at-risk” of becoming involved in terrorism. Despite the same top-down pressures and widespread public opposition to these policies, local governments and city-level communities diverge in their approaches to counter-extremism programs, ranging from active collaboration around these agendas to evasion. The existing scholarship lacks a theory to explain this puzzling sub-national variation. I argue that the structure of local Muslim civil society networks, in particular, the pre-existing relationships between Muslim civil society leaders, state actors, and other co-religionists, shape how local actors navigate steep trade-offs associated with these programs and ultimately respond to CVE agendas. Muslim civil society leaders, in particular, act as the causal mechanism that translates network structures into local policy outcomes, determining both whether bureaucrats decide to invest in counter-extremism policies and how communities take part in the long term. I test this theory in six cities over time, looking in-depth at CVE policy outcomes in four UK cities and two US cities, employing a mixed-methods approach that combines 139 interviews with local stakeholders and within-case social network analysis. The dissertation contributes to the literature on city governance and counter-terrorism, highlighting how sub-national community networks shape security policy outcomes, even amidst heavy top-down pressures and evolving terrorist threats.