Croatian WordNet is a lexical database created on the principles of the Princeton WordNet.As such, it contains a complex network of lexical relations built primarily on the relations of hyperonymy / hyponymy and synonymy.Thus WordNet can be regarded as a tool by which we can represent and in turn, closely examine various lexical-semantic relations of a language.However, in order to use WordNet as an analytical tool, it is necessary to provide a detailed account of other lexical relations such as polysemy, and also design the database according to the interactions of various lexical relations they have with each other.In this paper we strongly emphasize the role that WordNet can have in the study of complex lexical-semantic relations of a lexicon as a whole, and we try to show the methodological prerequisites that must be fulfilled in order to achieve this goal.We focus on the relations of synonymy and polysemy and their interaction in structuring the lexicon of the Croatian language.We show the similarities and differences between nominal and verbal polysemous structures and the ways these can be represented in the Croatian WordNet.We also show how certain methodological inconsistencies as pertaining to the original Princeton WordNet were corrected in the building of the Croatian WordNet.