This article applies aspects ofA Generative Theory of Tonal Musicby Fred Lerdahl and Ray Jackendoff to the analysis of Hindustani (North Indian) classical music, with a double-edged purpose. On the one hand, theGTTMmethodology is used to illuminate the workings of anālāp(as performed by Vijay Rajput, a vocalist in thekhyālstyle). On the other hand, the analysis acts as a case study to assess the viability of this methodology for the analysis of Indian classical music, and in particular, to test out Lerdahl and Jackendoff’s claim thatGTTMpresents a universal musical grammar, transcending specific cultures.This pilot study considers what modifications would be necessary toGTTM’s preference rules to make a viable generative theory for Hindustani classical music—or at least for its melodic aspects, governed as these are by the principles ofrāg. With such modifications, it is possible to formally represent levels of musical knowledge involved in the production and perception of this music, and to verify the generative principles whereby a performer is able to improvise a potentially infinite number of musical utterances from a finite set of rules.The investigation also fosters a critique of Lerdahl and Jackendoff’s theories. Among the questions addressed is that of the diachronic modeling of improvised musical performance—which, unlike the score-based studies ofGTTM, cannot be analyzed entirely outside the flow of time. Consistent with this critical position is the conclusion that, rather than the pursuit of universals per se, it is an openness to the tension between universals and particulars that may lead to the most valuable knowledge.