Duo musicians exhibit a broad variety of bodily gestures, but it is unclear how soloists’ and accompanists’ movements differ and to what extent they attract observers’ visual attention. In Experiment 1, seven musical duos’ body movements were tracked while they performed two pieces in two different conditions. In a congruent condition, soloist and accompanist behaved according to their expected musical roles; in an incongruent condition, the soloist behaved as accompanist and vice versa. Results revealed that behaving as soloist, regardless of the condition, led to more, smoother, and faster head and shoulder movements over a larger area than behaving as accompanist. Moreover, accompanists in the incongruent condition moved more than soloists in the congruent condition. In Experiment 2, observers watched videos of the duo performances with and without audio, while eye movements were tracked. Observers looked longer at musicians behaving as soloists compared to musicians behaving as accompanists, independent of their respective musical role. This suggests that visual attention was allocated to the most salient visuo-kinematic cues (i.e., expressive bodily gestures) rather than the most salient musical cues (i.e., the solo part). Findings are discussed regarding auditory-motor couplings and theories of motor control as well as auditory-visual integration and attention.