This paper argues that “improvisation” is not a sign of poor planning and failure to anticipate the features and consequences of a disaster, but is rather a recognition that “plans are hypotheses” about a projected future condition, as suggested by public-policy scholar Michael McGuire. Since the real conditions of a given disaster have distinctive features that differ from the most careful planning, adjustment to plans will always be necessary. Responses to disasters will inevitably involve dealing with the ambiguous and dynamic conditions of a given disaster that will require interpretation, judgment, and command-and-control decisionmaking related to real-world happenings, based on knowledge, training, practice, and experimentation derived from experience. This paper proposes a three-element typology of improvisational types. In “reproductive improvisation,” improvisers recreate an existing capacity; in “adaptive improvisation,” they amend an existing capacity in order to address changing demands, producing a new system; and in “creative improvisation,” they create an entirely new capacity in the absence of an existing model sufficient to address the current situation. Although the use of improvisation by a single individual in the course of a disaster can result in confusion and chaos in the response, when effective group improvisation occurs, the products of this collective problem solving can be effective responses to unusual situations. The skill of improvisation cannot be taught by assigning a script or following a checklist of steps. Its principles, however, can be taught, and the skill of improvisation can be developed through practices and exercises. Improvisation can be enhanced by removing organizational impediments to creative thinking and by creating an organizational culture that values improvisation. Examples of improvisation in the World Trade Center disaster are provided. 29 references