This article reformulates Hoekstra and Mulder's (1990) analysis of Locative Inversion in more current terms, while extending the empirical scope of their proposal to a larger set of constructions involving displacement of small clause predicates and VPs. The new proposal will stress H&M's basic claim that Locative Inversion in examples like Down the hill rolled the baby carriage is possible due to the fact that the predicative PP and the subject DP are in an agreement relation. We will adopt as an axiom that this agreement relation holds cross-categorially and resembles object agreement in that it involves agreement in w-features. If so, Hoekstra and Mulder's (1990) basic insight can be rephrased in more current terms as the following hypothesis: if A and B agree in w-features, both A and B can be the goal of some higher head H with unvalued w-features, and, consequently, be a candidate for internal merge with H. From this it follows as a corollary that in principle predicates may target any checking position that is normally targeted by the subject of the predicative phrase. This article will test this prediction by considering certain instances of predicate inversion in English, Dutch and Hungarian. #