It is common knowledge that specialised languages use some aspects of “languages” to express the communicative objectives of specialised communities. Compiling and analysing specialised corpora is one way of identifying recurrent features that tend to shed light on these special uses. As a result, it is possible to draw a list of the linguistic characteristics specific to a given specialised language, such as the use of a special lexicon or special lexico-grammatical features. Some of these features are so specific compared to other varieties of languages, whether specialised or not, that they will naturally be identified as typical of, say, legal English, economic English and so on. Therefore, lexis – and, to some extent, grammar – appears to be distinct from one specialised language to another, or from non-specialised languages to specialised languages, which would tend to show that specialised lexicons, together with some grammatical features, are a defining feature of specialised languages. Yet, as we will show, much of legal lexis is shared by other types of language, in particular everyday, non-specialised language, which implies that legal lexis is necessarily polysemous. Therefore the question is: how legal/specialised is legal lexis? Or, put differently: what makes legal lexis legal? Can legal lexis be considered as a characteristic of legal language? This article will first focus on some of the borrowings that characterise legal lexis in English and then look at some of the interpretation difficulties posed by a number of lexical units that do not stand out as being “legal” but yet consistently appear in legal messages and feature, for some of them, in legal dictionaries.