Every dictionary has to start with a list of headwords or dictionary forms. Selecting items for the list is a complicated issue on its own; however, even after the lexicographer arrives at the final selection of candidates, a new issue arises: how to deal with words that do not have a traditional dictionary form documented in the data (e.g. infinitive in verbs or masculine in adjectives)? Should the lexicographer reconstruct the unattested word form? In order to answer this question, addressed rather seldom and briefly in the literature (Wolski 1989, Schnorr 1991, Svensén 2009: 105–106; in Czech lexicography: cf. Čermák 1995, Filipec 1995, Kochová & Opavská 2016), this paper aims to describe the scope of this issue, to classify and interpret such headword lacunas from the perspective of Czech lexicography and suggest possible solutions based on large corpus data.