The present paper investigates semantics of human body part names (BPN) in languages of different structure. The lexemes under study are characterised by a high level of polysemy, frequent occurrence and primary role in the processes of world perception and categorisation. The empirical data comprise 438 lexemes (expressing 1438 meanings), which were analysed in seven languages of different structure (related and unrelated) from two language families: Indo-European (featuring such language groups as Germanic (English and German), Romance (Spanish), Balto-Slavonic (Latvian, Ukrainian and Russian)) and Japanese-Ryukyuan (featuring Japanese). The data are analysed within the framework of meronymy relation and highlight the universal features of semantic change in the compared languages, thus presenting a model for more extensive and complex studies of the semantic potential of basic lexical units. The research is based upon the methodology of classical semantics and typology as well as contrastive lexical studies, and is aimed at revealing the nature of semantic universals and regularities in the process of BPN semantic development. The lexical units under analysis are generalised and systematically investigated using contrastive analysis and linguistic description, whilst their semantic structure is thoroughly described with the help of componential analysis. The paper focuses on the classification of their regular semantic change mechanisms and juxtaposes the cross-linguistic similarities as well as differences in semantics of body part names. The relation of meronymy reveals interdependence with that of polysemy. The introduced index of polysemy helps to compare the potential for polysemy development in the languages under analysis. The conducted research enables the formulation of semantic universals and regularities within BPN polysemy development in languages of different structure as well as the creation of a semantic typology of BPN in related and unrelated languages. The analysis is supplemented with the statistical data allowing for the verification of the results. The main principles of this research may be applied to different thematic groups of lexis, contributing to further development of semantic language typology worldwide.