An emergency response process is a plan that outlines the workflow of different activities that need to be performed in response to an emergency situation. Effective response planning requires interoperability among systems belonging to different organizations. Unlike typical e-government processes that are well structured and have a well-defined outcome, emergency response processes are knowledge centric and their workflow structure and execution may evolve dynamically based on the environmental context as well as the type of service/activity invoked during process execution. It is impractical to define static plans and response process workflow for every possible situation due to the fact that unforeseeable situations may arise. Thus, a dynamic response requires adaptability to changing situation as the incident evolves. This article presents an approach for development of an Emergency Response Plan Recommendation and Composition System (ERPRCS) that, as the incident evolves, facilitates dynamic composition of an executable response plan by exploiting the knowledge of existing plans and learning from them. Specifically, the ERPRCS system generates an executable response plan from an existing response process that was instantiated for an incident most similar to the current incident. This response plan is customized based on available services and resource provider systems in a given jurisdiction and geographical area.