In this article I review two analyses of English existential constructions – the Case-transmission analysis and the expletive-replacement analysis – from the perspective of Chomsky's (1993, 1995) minimalist framework. I show that neither approach is free from empirical problems and that adopting minimalist program can reconcile these difficulties and account for more extensive data as well. I propose that LF feature-raising theory that I adopt here should be more economical than it is now, for example, in Chomsky (1995) so that only the relevant ‘checkable’ features raise at LF. My discussion concerns English existential constructions and Object Shift in Icelandic. In particular, I discuss (a) specificity, (b) scope and (c) negative polarity item licensing and show that the category-raising analysis of the associate of an expletive raises problems concerning these three aspects. The analysis presented in the article will be further supported by facts about predicate raising constructions.