The knowledge of the Portuguese lexicon has still many shortcomings. Some of them are well acknowledged; many other are quite unsuspected. Along the 20th century, Portuguese lexicography lost track of the ongoing research on word diachrony, morphological analysis and lexical semantics, unlike what happened with the lexicographic treatment of other comparable modern languages such as French, Castilian or Italian. Although Portuguese dictionaries generally include information about etymology and morphological structure, they present it quite inconsistently, since it is generally the output of the accumulation of what can be found in previous dictionaries. This modus operandi leaves no room to a systematic analysis of the set of words and word families that form each dictionary’s entry list. In fact, most contemporary dictionaries of Portuguese (paper editions and electronic versions as well) obey to conservative models, which tend to incorporate exhaustively the information made available by their predecessors, regardless of the real usage of the words. They devote a very considerable amount of effort to increase the entry list, accommodating neologisms (which are not that frequent) and specialized terms, randomly chosen by dictionary makers. From time to time, major dictionary publishers issue remakes of their own dictionaries for the sake of orthographic updates. Although users tend to be unaware of their deficiencies, this kind of dictionaries is far from being a useful working tool. In fact, since all previously registered words tend to occur in every new dictionary, without any mention to their usage, users are led to induce that they are equally available, and that is not the case. A different type of dictionary, with a critically selected word list and a thorough lexicographic description, is thus still lacking and still necessary. The project we will present in this paper, ROOTS, has been conceived as a further step to build such a dictionary.