ABSTRACT Field observation and of sound recording of Formosan macaques (Macaca cyclopis) at Mt. Longevity were conducted from Oct. 1999 to Oct. 2000. Spectrographic analyses revealed at least 25 basic patterns with 32 different vocal types from a total of 375 hrs recordings (55 hrs for scan sampling and 320 hrs for focal animal sampling). The three most frequent calls in overall vocal repertoire were coo calls (18.59%), hack (17.78%) and contact rattle (11.85%). The other calls included squeak (11.34%), noise and undulated scream (6.61%), sneeze (4.54%), greeting (3.77%), squeal (3.41%), vibrato growl (3.02%), growl (2.86%), cluck (2.54%), female copulation call (2.45%), squawk (2.14%), tonal scream (2.03%), alarm call (1.74%), threat rattle (1.66%), chuckle (0.92%), rise (0.61%), weeping (0.56%), male copulation call (0.32%), babble (0.30%), whine (0.29%), roar (0.27%), bark (0.24%) and mounting grunt (0.15%). Formosan macaques employ a complex vocal system composed of discrete signals as well as graded signals which vocal patterns connected by intermediate gradations. Discrete signal included coo call, greeting, contact rattle, mounting grunt, two copulation calls, babble, whine, weeping. Whereas the graded signals included the aggressive signals (threat rattle, growl, bark, roar, vibrato growl and alarm call), chuckle, the submissive signals (noise and undulated scream, squeal, tonal scream and squeak) and the distress signals (squawk, hack, cluck and sneeze). According to the context of emission, some of vocalizations could be divided into: (1) Affiliative contact calls: The calls, as the coo calls, greeting and contact rattles, may be summarized as affinitive contact calls for reduction and/or maintenance of close proximity between group members. (2) Aggressive calls: The growls, threat rattles and vibrato growls were uttered by dominant animals menacing sub-dominate group mates, members of other groups or other species. (3) Submissive calls: these submissive calls (including the various form of scream, squeal and squeak) with obviously structure differences. Male and female copulation calls and alarm call of M. cyclopis were clearly related to particular interactions or specific external stimulus. The acoustic structures might be influenced by specific social factors, motivation or the arousal state of the callers. The acoustic structure of submissive calls was associated with the aggression with or without physical contact. Noise and undulated scream were usually used in aggressive interactions with physical contact (68%), whereas squeal, tonal scream and two types of squeak were used without physical contact (75%). Age specific difference of vocal behavior seemed to more pronounced than asymmetries between the sexes. Infants have the highest vocal rate (1142.2 vocalizations/10 h) and relative frequency of vocalization (44.58 %) than other age/sex classes and they decreased from infants to adults. With increasing age, visual signals could not only complement vocal signals but also replace the vocal cues. The relative frequency of four vocal patterns (coo call, greeting, squeak and hack) decreased from infants to adults, whereas that of threat rattle and growl increased from infant to adults. On the other hand, there were five vocal patterns (including cluck, sneeze, whine, weeping and babble) only used by juveniles and infants. In addition, seven vocal patterns included squeak (64 %), hack (63.6 %), squawk (76%), cluck (85.7 %), sneeze (90 %), weeping (88 %) and babble (100%) mainly were used by infants. Among them, three vocal patterns (squeak, hack and squawk) were produced mostly by infant II. Only infants used babble. Roar and bark were only present in adults and sub-adults. Obvious sex differences of relative frequencies of vocal patterns occurred in adults, the vocal activity of females consistently higher compared to that of males. In six patterns (vibrato growl, chuckle contact rattle, squeal, squeak and hack), females uttered more often than male peers. Mounting grunts and male copulation calls were exclusively used by adult males whereas female copulation calls were only produced by adult/sub-adult females. Vocalizations were produced/ceased by certain sex/age class that might associate with social organization, morphological feature and circumstances experience. The vocal repertoires of Formosan macaques revealed the high similarities in the species within genus Macaca, especially to M. fuscata, M. mulatta and M. radiata. It may be due to phylogeny, habitats and social organizations.