The democratic boundary problem arises because it appears that the units within which democratic decision procedures will operate cannot themselves be constituted democratically. The study argues that setting the boundaries of democracy involves attending simultaneously to three variables: domain (where and to whom do decisions apply), constituency (who is entitled to be included in the deciding body) and scope (which issues should be on the decision agenda). Most of the existing literature has focussed narrowly on the constituency question, endorsing either (...) the All‐Affected Interests Principle or the All Subjected Principle, but neither is satisfactory as a general solution. In particular, the former fails to explain why having interests at stake in a decision necessarily gives you the right to participate in making it, and the latter, although more plausible on that count, assumes that the domain and scope issues have already been settled. To make progress, we need to bring democratic values to bear on the boundary problem. The units we favour should be those that are likely to promote political equality and solidarity among members of the demos. Although this approach will often justify existing territorial states as sites of democracy, it can also generate arguments for making boundary changes along one or other dimension. (shrink)