Using functional magnetic resonance imaging, the distribution of hemodynamic brain responses bound to the perceptual processing of interjections, that is ‘exclamations inserted into an utterance without grammatical connection to it’, was determined (vs. a silent baseline condition). These utterances convey information about a speaker's affective/emotional state by their ‘tone’ (emotional prosody) and/or their lexical content. Both communicative aspects of interjections elicited significant bilateral blood-oxygen-level-dependent signal changes within superior temporal cortex. In addition, affective-prosodic cues yielded hemodynamic activation of the posterior insula as well as cortical/subcortical structures engaged in the control of innate emotional behavior. These observations corroborate the suggestion that interjections might trace back to proto-speech vocalizations of an early stage of language evolution.