The vocal repertoires of nonhuman primates have long been thought to be invariable across populations and not to result from vocal learning. However, increasing evidence suggests that learning does influence vocal production in nonhuman primates, and that several species modify the structure of their calls in response to social or environmental influences. Vocal usage learning refers to the process whereby an individual learns in which circumstances to produce a certain call type, whereas vocal production learning refers to the process in which signals get modified as the result of individual experiences. Common marmosets (Callithrix jacchus) show socially mediated vocal plasticity as adults and during vocal development. This propensity to engage in simple forms of vocal production learning (accommodation) should produce population-level differences in call structure. To test this prediction, we compared the vocalizations of three captive populations of common marmosets. We analyzed the acoustic structure of 1337 phee calls, 461 trills, and 3611 food calls and compared them with a permutated discriminant function analysis. We found that all call types differed significantly between the three populations, and 76–98% of the calls were correctly classified. As physical differences in body mass and environmental differences between colonies could not explain the call differences, we conclude that vocal accommodation is the most likely explanation for the differences in call structure. This will allow us to further investigate the role and importance of vocal learning in a species increasingly used to study vocal learning and language evolution.