This thesis focuses on lexical collocations in the language of business and analyses them within the fields of phraseology, lexicography, corpus linguistics, and glottodidactics. Collocations are examined as multi-word units that have a fixed structure with a limited commutability of constituent words. Collocations are considered to be binary units in which both constituents are content words. An indispensable part of collocations, which are necessary for their language accuracy, are also function words, e.g., articles in English. A collocation can also be expanded by merging two or more collocations, thus creating complex collocations. The binary structure is considered practical from a lexicographic point of view. It is applied in the qualitative and quantitative analysis of the available monolingual English dictionaries and bilingual English-Croatian and Croatian-English dictionaries of the language of business, which was conducted to determine the methods of presenting collocations in dictionaries and the level of their accessibility to end-users. The analysis revealed that collocations are regularly a part of dictionary entries in monolingual British dictionaries, but not in monolingual American and bilingual English-Croatian and Croatian-English dictionaries. The research into dictionaries was followed by an analysis of the Business English Corpus and the Business Croatian Corpus with the aim of creating a small database of lexical collocations in the language of business that could be used to improve the existing bilingual dictionaries or to write new business English-Croatian and Croatian-English dictionaries. The lexicographic and dual corpus analysis results in a model of a dictionary entry for bilingual LSP dictionaries, which includes basic grammatical information about the headword, collocations with the headword in the source language, translations of collocations in the target language, and examples of collocation in context.