Acoustic signals are expressions of phenotypic diversity and their variation could provide important information on differentiation patterns within species. Due to a number of selective pressures acting on signals, the contribution of genetic drift is often complex to outline. This study aims at evaluating if an examination of the acoustic structure of communication signals can allow the identification of evolutionary units in species capable of vocal learning. We quantified and compared parameters of whistles emitted by three dolphin species (Stenella coeruleoalba, Delphinus delphis, and Tursiops truncatus) to examine the hypothesis that acoustic signals permit the recognition of differentiation between populations from the Atlantic Ocean and the Mediterranean Sea. In the three species, recordings were correctly assigned to their basin of origin with a percentage higher than 82% by DFA. Frequency parameters were the most stable within each species. Where gene flow has been shown, i.e., within Atlantic Ocean, significant differences were found principally in modulation parameters. We hypothesize that these parameters are influenced by social and behavioral factors and that similar ecological conditions led to convergent acoustic features. Results of this study suggest that is it possible to recognize evolutionary units based on acoustic data.