Alarm and estrous calls emitted by Japanese macaques were recorded and analyzed in the Arashiyama West and East groups. Their responses to natural calls as well as to synthesized versions varying in the acoustic parameters that defined the vocalizations were studied. The response patterns shown by Arashiyama West group members, which were subject to a distinct change with only a slight difference of a single parameter, appeared to reflect strict underlying perceptual boundaries. This was analogous to the categorical perception that humans show with speech sounds. In contrast, continuous perception was exhibited by Arashiyama East group individuals. When several sounds were played back in combination to the former group, following stimuli were recognized by quite different cues from those by which the first sound was perceived. The groups' differences in vocal perception are discussed in terms of the ecological differences of the environments they inhabit.