We study pandemic policy in a global game. Independent regions face the outbreak of a disease and can exert effort to control it. If the disease is infectious, it will be controlled only if sufficiently many regions exert effort. Regions thus face a coordination game which, in the well-studied case of perfect information, has multiple Nash equilibria. We show that even a vanishing amount of uncertainty about fundamentals of the game forces selection of a unique equilibrium. In well-identified cases, a pandemic occurs even though it is inefficient and could be avoided. Harmful diseases are less likely to become an pandemic while diseases which require greater cooperation have a larger chance to go uncontrolled. A higher cost of effort increases the probability of a pandemic and regions whose actions exhibit stronger spillovers have a greater influence on the limiting equilibrium. Given the possibility of an inefficient but rational pandemic, we also study a mechanism to facilitate coordination on disease control. In particular, we introduce the concept of P-delegation, a kind of conditional delegation which binds delegating regions to exert effort toward disease control if and only if at least a weighted proportion P of all regions has delegated. We show that, for judiciously chosen P, P-delegation can help avoid pandemics. Our results thus suggest a way forward for the international cooperation on disease control.