Research in musical performance studies has generated a healthy scepticism of the importance of large-scale structure to performance (in terms of both interpretation and perception): on the one hand, it might well be hardwired into notation, and on the other, prioritising it risks simply repeating outworn maxims that neglect the performer’s musical contributions. Recently, some scholars have begun to rethink the potential structural relevance of performance rather than necessarily determining structure on the basis of the musical score alone. In this article, I consolidate some of this thinking and draw out its implications for performers’ handling of large-scale structure; in doing so, I suggest that we should consider moving away from conventional large-scale score-based forms as structural certainties. I support this through a case study of Johannes Brahms’s Intermezzo in E minor, Op. 119 No. 2 in which I analyse recorded performances by Wilhelm Backhaus, Maria Yudina and Ilona Eibenschütz. I conclude by arguing that the inclusion and prioritisation of any particular musical material – whether the score, performance, or other – requires serious consideration and reflection in any analytical act.