This paper tries to show that although Carnap's distinction between internal and external questions in terms of a linguistic framework is philosophically important, and that although metaphysical questions are, as Carnap claims, external questions, Carnap's conclusion that all meaningful metaphysical questions are practical questions about language is not justified. This is done in three steps. First, it is argued that it is plausible to suppose that there is for languages a kind of external question other than the one kind Carnap specifies, because “language games” are like the game of chess in important ways and there seems to be such a kind of question for chess. Second, it is shown that at least some metaphysical questions can quite reasonably be interpreted as being of this kind. Third, reasons are given for rejecting Carnap's grounds for claiming that there is only one kind of external question.