Over the past decade in Latin America, there have been several cases of attempted and successful closures of congress through the mechanisms of direct democracy and constitutional reform. This is qualitatively different than previous closures through military coups or autogolpes. When are presidents more likely to attempt to close congress by this means and what factors determine whether it will be successful? I argue that several factors make a president more likely to attempt this strategy: a radical agenda and a minority in congress (incentives), perceived chance of success in relation to alternative strategies, and diffusion effects. In turn, success is determined by the presence of two factors. One is mobilizational leverage, which I measure by examining the presence of an electoral mandate, high presidential approval, and the ability to rally large sectors of the electorate around the president's agenda. The second is institutional leverage, which I measure by examining party system weakness, and the neutralization of non-legislative institutions, such as the military, Supreme Court, and electoral council. I test this theory in four case studies: Venezuela, Ecuador, Honduras, and Nicaragua.