This paper presents a micro-temporal analysis of a complete, 34-minute long vilambit (i.e., slow) sitar and tabla duo performance by Pt. Nikhil Banerjee and Zamir Ahmed Khan. The analysis involved extracting the onsets of approximately 3,000 musical sound events—the deliberate manual strikes on the sitar strings and tabla skins—produced on the metric level of ‘maximal pulse salience’ (i.e. half-mātrā) in the course of performance. These data were used to analyze performers’ tempo fluctuations, isochronization, mean asynchronization, and interpersonal entrainment throughout the vilambit and across various formal sub-sections. The analysis reported here is original in that it tackles an extended duration of a North Indian instrumental duo performance, with the aim of relating micro-temporal patterns to other variables such as alternating musical roles, surface rhythm, metric cycle, and formal development. Results reveal consistent patterns of micro-temporal interaction, including: (i) linear micro-fluctuations in tempo relating to the performers’ alternating roles as soloist and accompanist; (ii) cyclical micro-fluctuations in tempo according to metric framework; (iii) higher isochronization values when performers act as soloists; (iv) higher mean asynchronization values during tabla solos; (v) persistent phase relationships according to musical role; (vi) marked changes in synchronization and coupling according to surface rhythm and formal development; and (vii) cyclical patterns in levels of synchronization, phase, and coupling relating to metric framework. The study of sitar and tabla performers’ micro-temporal interactions thus exposes a new functional level in which musical roles, expression, and formal development are enacted in this genre.