In this article, an argument is made for representing theta-positions structurally through a UTAH-like mapping onto articulated VP structures. Further, an argument is made for giving these positions a distinct status. It is argued, using data from binding in Balinese and Malagasy, that binding may be sensitive to theta-positions (T-positions), which are a subset of A-positions. Specs of lexical categories must be T-positions, Specs of functional categories cannot be T-positions. It is argued further that, once it can be shown that binding may be sensitive to T-positions, one is forced to revise what can count as a T-position in order to account for raising constructions. Categories which bind event variables (such as Aspect) are seen to be non-distinct from lexical categories and from functional categories. As such, their Specs may count as T-positions, explaining why an NP in a derived position may act as an antecedent.