Domestic mobility restrictions to control the spread of COVID-19 are widespread in developing countries, and have trapped millions of migrant workers in hotspot cities. We show that bans can increase cumulative infections relative to a counterfactual san restrictions. A SEIR model shows bans’ impacts are nonlinear in duration. We empirically test this hypothesis using a natural experiment in India as well as data from China, Indonesia, the Philippines, South Africa, and Kenya. Although very short and long restrictions limit the spread of disease, moderately lengthy restrictions substantially increase infections. This suggests significant caution before imposing mobility restrictions in developing countries.