This paper investigates the Possessive Dative Construction (PDC) in Hebrew and Romance, and centers on the puzzling nature of the Possessor Dative (PD) - a semantic argument of the possessee which behaves like a syntactic argument of the verb. A variety of structural tests indicate that the possessee contains an empty category bound by the possessor, as previous researchers have concluded (Guéron, 1985; Borer and Grodzinsky, 1986); however, contrary to what the standard 'thematic' analysis of PDC maintains, it is argued that this is a genuine movement dependency. A case-driven possessor-raising account is developed, which explains the possessor-possessee co-occurrence restriction, and the interaction of PDC with extraction and control phenomena. The claim that PD raises to a specifier position projected by the verb further derives a significant cross-linguistic generalization - namely, that PDC is incompatible with non-agentive dyadic verbs. This generalization supersedes the 'theme-affectedness' condition, which is shown to be empirically false. The syntax of PDC proves a useful tool to probe into the structure of VP across various verb classes.