We examine three knowledge-based processes and interactions underlying an effective improvisation capability in research and development (R&D) teams: a team’s ability to create a shared understanding of new knowledge, a team’s experience working together, and a team’s ability to gather external knowledge. Using a sample of 100 R&D teams developing computer technology innovations, we also examine the moderating role of "minimal structures" (goal clarity combined with autonomy) as a contextual factor supporting effective improvisation. Our results detected different ways in which the processes and interactions involved in an improvisation capability interacted with the context. We find support for the positive relationship between shared understanding of new knowledge and improvisation capability and find that this relationship is strengthened by minimal structures. Team’s experience working together was not associated with improvisation capability, but when minimal structures are present, more experience working together is positively related to improvisation. Finally, a team’s external knowledge-gathering ability is positively associated with improvisation, but, surprisingly, when minimal structures are present, this positive effect is reduced. We conclude with implications for improvisation theory and for the practice of R&D teams.