This chapter discusses the role of the most-similar systems design and most-different systems design in comparative public policy analysis. The methods are based on John Stuart Mill’s ‘eliminative methods of induction’ and regularly mentioned in educational books in social research methodology in general and in comparative politics in particular. However, their role has been less prominent in the literature on public policy. The methods are particularly useful in cross-country studies where the number of research units is limited, but where the researcher operates with a variable-oriented approach. The present contribution outlines how the most-similar systems design and the most-different systems design should be applied in deductive and inductive research efforts. It also challenges claims that the methods are unable to cope with interaction effects and multiple causation. The chapter also discusses to what extent the two methods are still relevant, given the recent advances made in multi-level modelling and in qualitative comparative analysis (OCA). Here, the argument is raised that the most-similar systems design and the most-different systems design are useful for accounting for complex social processes, and also have the advantage of making it possible to eliminate a large number of potentially relevant explanatory variables from further analysis. As a consequence thereof, regression models using multi-level techniques should often be constructed according to the logical foundations of a most-different systems approach in situations where the number of cases is low at the system level.