In this article, I examine the repeated musical formulas that appear in ṭhumrī performances. Chunks of recurring musical material, my analysis has revealed a large number of different types of formulas in ṭhumrī recordings by a wide variety of musicians from throughout the twentieth century. Here, I propose ways of understanding how and why they occur in ṭhumrī and suggest ways of taking them into account in an analysis of ṭhumrī style. In Parts I and II, I lay some of the theoretical groundwork for my analysis of formulas in ṭhumrī. In Part I, I consider the nature of improvisation in North Indian classical music and begin to explore the significance of musical formulas in ṭhumrī. Attempting to account for their widespread occurrence, I argue that they are a result of the way in which North Indian classical musicians teach, learn and practice ṭhumrī, in preparation for largely improvised performances. In Part II, I explore wider contexts in which to make sense of formulas in North Indian classical music; specifically, I draw attention to parallels with the formulas that appear in other musical traditions, oral poetry and spoken language. In Part III, I take the formulas of ṭhumrī as a starting point for an analysis of ṭhumrī style. Categorizing these formulas into different types, I distinguish, for example, between the precise repetition of memorized musical figures and the repeated use of abstract musical strategies, which produce entirely different musical phrases whenever they appear. I consider the role that formulas play in delineating phrase structure and creating a sense of musical syntax. Finally, I argue that the examination of formulas offers a means of characterizing different ṭhumrī styles.