During many crises, access to sensitive emergency-support information is required to save lives and property. For example, for effective evacuations first responders need the names and addresses of non-ambulatory residents. Yet, currently, access to such information may not be possible because government policy makers and third-party data providers lack confidence that today's IT systems will protect their data. Our approach to the management of emergency information provides first responders with temporary, transient access to sensitive information, and ensures that the information is revoked after the emergency. The following contributions are presented: a systematic analysis of the basic forms of trusted communication supported by the architecture; a comprehensive method for secure, distributed emergency state management; a method to allow a userspace application to securely display data; a multifaceted system analysis of the confinement of emergency information and the secure and complete revocation of access to that information at the closure of an emergency.