Policy diffusion is a common phenomenon in federal states: indeed, one of the normative justifications of decentralized policy making is that it permits the development and spread of best practices. Following Berry and Berry (1990), event-history analysis has been the method of choice for the quantitative investigation of policy diffusion, but Volden (2006) has recently introduced a dyadic variant of this method in which units of analysis are not states but, instead, pairs of states. This article discusses the dyadic approach with a particular focus on the diffusion of policies in Switzerland. The goal is not to introduce a new method, but rather to provide a practical overview for researchers interested in using it. The article shows how the method has migrated from the international relations literature to the policy-diffusion literature, describes the typical structure of a dyadic dataset in a diffusion context, and discusses several modeling issues. The usefulness of the dyadic approach is illustrated empirically with the example of health-insurance subsidy policies in Swiss cantons.