In this paper, we propose a secure distribution model for multimedia surveillance data, where different locations of the monitored facilities may have different security requirements. We propose a multilevel security framework, wherein the surveillance data streams are classified according to the sensitivity of the location of the surveillance devices, and users (guards) have corresponding security clearances. Guards monitor live multimedia feeds emanating from surveillance devices placed throughout the facility. Our position is that during normal mode of operation, guards are allowed to access only those multimedia streams for which they have the proper authorizations. However, in an emergency, guards may receive pre-computed emergency instructions and/or cover stories. We show how to compose multilevel secure SMIL documents, compute views for each security classification, enforce integrity, confidentiality and access control, and deliver the secure views to handheld devices while 1.