This study investigates passive and impersonal constructions in three Romance dialects of Italy: Sicilian, Abruzzese and Tuscan. The aim of this investigation is twofold: one the one hand I aim to contribute to the documentation of the dialects of Italy, and on the other, I aim to build on existing theories on passive and impersonal constructions. Whilst there is to date no detailed discussion of passive structures in Italo- Romance dialects, works such as Cennamo (1997), Ledgeway (2009) and Rohlfs (1969) suggest that passive structures are unpopular in the dialects of Italy. I explore the extent to which the passive is rejected in each dialect, and, where it is not used, which constructions are used in its place. From the data I have collected on passive constructions in these three dialects, a pattern emerges: the acceptability of the passive directly relates to the transitivity of the verb. Whilst the tolerance of the passive varies from dialect to dialect, one consistent result is that passives are affected by the ranking of the verb on the Transitivity Hierarchy (Hopper and Thompson 1980) and its entailments. This thesis presents interesting findings on si impersonals, uno, third person plural impersonals and the Abruzzese nome, which of particular interest, and which I claim is an impersonal clitic, that holds a plural feature. Throughout my discussion of impersonals, I develop a hierarchy of impersonals, which is based on the semantic features (+/- referential, +/- definite, +/- irrealis) of each impersonal type. This hierarchy captures all of the impersonal types found in the three dialects, and allows for cross-dialectal analysis. I present novel findings relating to the Tuscan first person plural impersonal si and the split in its usage, which, to my knowledge has not yet been discussed in the relevant literature. The data show that Tuscan first person plural impersonal si can be used with transitive and unergative verbs but not with unaccusatives. I propose a reason for this split, which is based on what I term agreement features (person, number, gender). As well as providing a detailed analysis of each impersonal type, I attempt to refine existing definitions of impersonals and propose that all types of impersonal constructions can be defined by their deficiency in one of the three agreement features, in other words, they are âfeature deficient constructions'.