The presented study aimed to revealing the possibility of human recognition and automatic classification of the emotional states by vocalizations of rhesus macaques. This work is a continuation of the study of emotional states manifestation in vocalizations of primates. The participants of the study were 9 unrelated rhesus macaques, aged 14 - 36 months. 10 adult people (5 specialists working with macaques and 5 listeners – physiologists with professional experience in the field of child speech) took part in the perceptual study, classified emotional states by vocalizations of macaques. The vocalizations of animals were collected in laboratory condition and annotated to emotional states: “joy-neutral state-sadness-anger-fear”. The original dataset of emotional vocalizations of rhesus macaques was created. Perceptual analysis revealed that specialists could recognize all emotional states by vocalizations of macaques, with maximal accuracy for sadness, minimal accuracy – for joy state; listeners recognized emotional states worse vs specialists. In automatic classification, we obtained the Unweighted Average Recall (UAR) value close to UAR value for recognition by specialists. Fear, joy, and sadness states were automatically classified better than anger and neutral states. The data obtained in the study of rhesus macaques’ vocalizations could be useful for comparison with data on human and nonhuman primates. In practice, the information about features of affective calls of macaques is important for specialists working with animals in laboratory conditions.