If the commands of authority are peremptory and content-independent directives, it is a great puzzle why any rational autonomous agent should accept them as morally binding, as Robert Paul Wolff and others have argued. I analyse the peremptory and content-independent quality of authoritative directives and argue that all earthly authorities operate within a specified domain. I investigate three candidates for the role of universally applicable boundary conditions–morality, harm to self, and absurdity. I conclude that commands are authoritative only when intra vires, i.e. issued within the proper domain of the authority. Wolff's challenge is not met, hut it is shown to be less forbidding.