Why is there no NATO in Asia? Literature on this question is selective and incomplete. This paper develops a new theory with determinate predictions regarding patron and clients’ alliance design preferences, the alliances that result, the commitments therein, and alliance duration. A subtle but nonetheless persistent form of entrapment problem exists with clients that don’t want a war yet fear adversary aggression. Clients’ commitment to collective security is a hand-tying costly signal that assures the patron of client resolve to defend the status quo and reduces the probability and costs of entrapment. Patrons will rationally prefer to join in an alliance clients who have already made a collective security commitment. Clients are more likely, the paper shows, to make such commitments when their adversary credibly threatens to militarily occupy at least one of them. This is more likely in land than in sea theatres. When clients fail to realise collective security, patron efforts to impose it on them will fail, will result in short lived multilateralism, and will force bilateralism on the patron. The paper uses new archival evidence from Britain and Australia to show how this strategic framework explains variation in alliance design in Europe and Asia.