I have reviewed in this chapter a variety of truthconditional approaches to the semantics of questions. They exhibit a sort of progression in terms of empirical coverage. The categorial approaches have prima facie difficulty with coordination and the associated entailments, as well as with embedding, because of the nonuniformity of the semantic types associated with questions. The Aqvist/Hintikka approach handles coordination and certain cases of embedding, but has difficulties with other cases, involving verbs such as wonder and ask. The propositional approaches deal with these, but at least those of Karttunen and G&S, which make questions denote one true complete answer at an index, lose the universal/existential ambiguity captured handily on the Aqvist/Hintikka account. G&S, it is true, introduce additional meaning rules to capture this ambiguity, but once it is admitted that questions can have interpretations on which they don't have just one true complete answer, the question arises whether it is possible to give an adequate analysis according to which this is the general interpretation of questions, in effect, a return to the intuitions behind both the Aqvist/Hintikka and the Hamblin accounts, but overcoming some of their inadequacies. In the rest of this study, this is what I attempt to do. My analysis consists of a generalization of the Aqvist/Hintikka approach for certain question-constructions, plus a version of the Hamblin approach for others, carried out within the basic assumptions of the LHK theory of quantification summarized in chapter I. A consequence of my analysis is that the class of question constructions divides into two logicosemantically distinct types, and much of the empirical effort of this study is directed towards supporting this division.