To what extent and in what arenas do collaborating musicians need to understand what they are doing in the same way? Two experienced jazz musicians who had never previously played together played three improvisations on a jazz standard (“It Could Happen to You”) on either side of a visual barrier. They were then immediately interviewed separately about the performances, their musical intentions, and their judgments of their partner's musical intentions, both from memory and prompted with the audiorecordings of the performances. Statements from both (audiorecorded) interviews as well as statements from an expert listener were extracted and anonymized. Two months later, the performers listened to the recordings and rated the extent to which they endorsed each statement. Performers endorsed statements they themselves had generated more often than statements by their performing partner and the expert listener; their overall level of agreement with each other was greater than chance but moderate to low, with disagreements about the quality of one of the performances and about who was responsible for it. The quality of the performances combined with the disparities in agreement suggest that, at least in this case study, fully shared understanding of what happened is not essential for successful improvisation. The fact that the performers endorsed an expert listener's statements more than their partner's argues against a simple notion that performers' interpretations are always privileged relative to an outsider's.