This paper considers the nature of compounding in Sakha (Yakut), a Turkic language spoken in north-eastern Siberia (Russian Federation), and provides both descriptive and theoretical grounds for classifying nominal compounds in Sakha. The focus of the study is on primary compounds and therefore synthetic, deverbal compounds are left out of consideration. Arguments are offered for a mixed classification of nominal compounds in Sakha based on both semantic and formal features. The main reason for the necessity of a mixed approach lies in the discrepancy between semantic and formal characteristics of a compound. Although a compound can be endocentric (having a head) from a formal point of view, it can nevertheless be exocentric (without a head) on semantic grounds. The following three types of nominal compounds will be distinguished – root compounds, possessive compounds and dvandva compounds. It will be argued that even though a compound has a complex syntactic makeup consisting of more than one word, nevertheless it functions as a single lexeme/word and also denotes as such.