This dissertation investigates the concept of finiteness across Romance, a theoretical notion very frequently used within linguistics although still poorly understood (Ledgeway 2007:335). Various Romance verb forms and clauses which are not readily classifiable as either finite or non-finite are examined, such as personal and inflected infinitives, as well as Balkan-style subjunctives in Romanian, Salentino, and southern Calabrian, morphologically finite verb forms which behave syntactically as non-finite verbs. First the categorial status of irrealis complements is studied; it is argued that both non-finite (the Romance infinitival complementisers deriving from Latin AD and DE) and irrealis complementisers (southern Calabrian mu, Salentino cu, Upper southern Italian che and Romanian să) are spurious categories. Specifically, it is shown that AD, DE, mu and cu can head variously sized clauses with different degrees of syntactic finiteness and that the morphological form of the verb does not seem to influence the clause size nor the degree of finiteness. Romanian să-clauses, on the other hand, are consistently CPs. It is thus concluded that there is no cross-linguistic correlation between finiteness and clause size. Second, the diachrony of these irrealis complementisers is analysed as well; they all result from a process of downwards (re)grammaticalisation, whereby grammatical elements originating in the C-domain come to occupy the lowest position of the CP, and, in the case of mu and cu, also come to head smaller complements and thus to occupy lower functional heads. This process is accompanied by a reanalysis from phrase (XP) to head (X) with concomitant phonological reduction. Third, it is shown that, unlike finite verbs, non-finite and semi-finite verb forms consistently move to a high position within the clause. This is also true of subjunctives; all have a common feature that requires the verb be moved to the edge of the inflectional domain. This movement renders the regular subject position SpecTP unavailable in most of these cases. The central proposal of the dissertation is that finiteness is not a linguistic primitive, but should be broken down into the anchoring of both Tense and Person. Both allow for different degrees of anchoring to the speech act (independent, dependent, or absent). There is an asymmetrical relationship between the two: only when Tense anchoring takes place, can Person anchoring obtain too. The combination of both anchoring mechanisms yields a scalar view of finiteness that matches more closely the wide range of semi-finite and non-finite forms explored in the dissertation. It is the dependent anchoring which triggers non-finite and semi-finite verbs to move high, while the absence of this anchoring automatically renders reduced complements non-finite. Finally, only when both anchoring mechanisms act completely independently does a fully finite clause obtain.