Existing research has shown a link between social identification and motor coordination: identification affects performance of coordinated action, for example. However, beyond research focusing specifically on synchronous movements, there is no systematic approach examining more general motor coordination as a causal factor in social identification. Drawing on joint action theory, we predicted that any form of coordination should induce co-representation, which should in turn increase social identification. We tested this in a novel dual-joystick paradigm in which participants performed joint tasks varying in (non-synchronous) coordination requirementsin a repeated-measures design (N=45). When task segments were discrete between partners, they identified less as a group than when they had to coordinate their behavior. Furthermore and surprisingly, although constant coordination increased co-representation relative to intermittent coordination, it did not correspondingly increase social identification. Exploratory analyses showed that worse performance in the constant coordination task may explain this finding, as performance correlated positively with social identification (only) in the coordination conditions. A second experiment tested this explanation by subtly manipulaing performance in otherwise identical constant coordination tasks in a repeated-measures design (N=53). In line with the preceding results, social identification was higher in the high performance condition, while co-representation was the same regardless of performance. Our findings suggest both that coordination in general may increase social identification and that effective coordination performance is an important contributing factor. Thus, we provide impetus towards integrating existing findings that synchronous movement effects social identification into a broader theoretical perspective informed by joint action theory.