Motor simulation has been implicated in how musicians anticipate the rhythm of another musician's action to achieve interpersonal synchronization. Here, we investigated whether similar mechanisms govern a related form of rhythmic action: dance. We examined (1) whether synchronization with visual dance stimuli was influenced by movement agency, (2) whether music training modulated simulation efficiency, and (3) what cues were relevant for simulating the dance rhythm. Participants were first recorded dancing the basic Charleston steps paced by a metronome, and later in a synchronization task they tapped to the rhythm of their own point-light dance stimuli, stimuli of another physically matched participant or one matched in movement kinematics, and a quantitative average across individuals. Results indicated that, while there was no overall "self advantage" and synchronization was generally most stable with the least variable (averaged) stimuli, motor simulation was driven-indicated by high tap-beat variability correlations-by familiar movement kinematics rather than morphological features. Furthermore, music training facilitated simulation, such that musicians outperformed non-musicians when synchronizing with others' movements but not with their own movements. These findings support action simulation as underlying synchronization in dance, linking action observation and rhythm processing in a common motor framework.