Very specific vocalizations (girneys) are interchanged extensively between Japanese monkeys before they establish grooming contact. I undertook acoustic classification of vocalizations tape-recorded in a captive group, and found that they could be divided into two broad classes according to the position of the peak of the tonal element relative to the whole call length. Sequelae analyses revealed that the two variants were linked with different probabilities of the caller's subsequent behaviour. When the peak occurred in the first third of the girneys their callers were likely to groom the signal recipients, while when the peak occurred in the final third of the calls, their vocalizers tended to receive subsequent grooming. Playback experiments showed that the animals really perceived those variants. Japanese monkey contact call variants have specific and distinctive motivational referents.