While most discussions of Ramsey conditionals and most “solutions” to Gardenfors’ challenge assume a purely syntactic, and purely propositional, perspective (following in this sense Gardenfors himself and the original AGM approach), in this paper we adopt a semantic, and more specifically modal, point of view. While Ramsey, Gardenfors and others only give conditions for “acceptance” (i.e. belief by an implicit agent) of a Ramsey conditional, we are primarily interested in its truth conditions. Indeed, from a semantic and modal p erspective, it makes no sense to ask for belief (or acceptance) of a statement whose extensional meaning was not yet defined. In this sense, Ramsey’s test cannot be taken in itself as a “definition” of a Ramsey con- ditional A > B, since it does not specify in what possible worlds might the conditional be true. So, in a semantic-modal setting, the important question is: are there any truth conditions for A > B that would be compatible with Ramsey’s test (given the usual modal semantics for belief, and given some reasonable semantics for belief revision)?