Past research has shown that mimicry has a number of pro-social consequences for interaction partners. However, such research has almost exclusively focused on its effects among interaction dyads. As social interactions are often witnessed by third-party observers, the question arises which inferences perceivers draw from observing mimicry. In the present work, we apply a third-party perspective to mimicry and test whether observers perceive mimicking individuals as submissive. Experiment 1 confirmed our prediction and found that observers perceived a mimicking person as less dominant, and thus more submissive, than a mimicked person. Experiment 2 replicated this finding and ruled out possible demand effects. Experiment 3 showed that when an interaction partner does not mimic the movements initiated by another person, the interaction partner gains dominance in the eye of the observer. Experiment 4 demonstrates that the inferences that perceivers draw from observing mimicry partly rely on a mere action-response pattern. These findings have not only important implications for mimicry as a genuinely social phenomenon, but also for research on impression management and person perception.