Previous research has demonstrated perceptual aftereffects for emotionally expressive faces, but the extent to which they can also be obtained in a different modality is unknown. In two experiments we show for the first time that adaptation to affective, non-linguistic vocalisations elicits significant auditory aftereffects. Adaptation to angry vocalisations caused voices drawn from an anger-fear morphed continuum to be perceived as less angry and more fearful, while adaptation to fearful vocalisations elicited opposite aftereffects (Experiment 1). We then tested the link between these aftereffects and the underlying acoustics by using caricatured adaptors. Although caricatures exaggerated the acoustical and affective properties of the vocalisations, the caricatured adaptors resulted in aftereffects which were comparable to those obtained with natural vocalisations (Experiment 2). Our findings suggest that these aftereffects cannot be solely explained by low-level adaptation to acoustical characteristics of the adaptors but are likely to depend on higher-level adaptation of neural representations of vocal affect.