Interest in federal institutions has grown rapidly over the past two decades, producing a voluminous literature. While a variety of interesting questions underlie this research, the bulk of it has been driven by the resurgence in ethno nationalism in both developing and developed states. No multiethnic federal democracy has yet collapsed, but several nd themselves under increasing strain from ethnic elites and movements demanding greater autonomy or outright independence. As early as 1977, participants in a Publius symposium warned governments about the centrifugal forces that federalization might unleash (Duchacek, 1977). Despite such warnings and dangers, however, it is surely no exaggeration to say that today many scholars rush to endorse federalizing strategies for dealing with ethnic and other disintegrative forces (cf. Horowitz, 1985;Cohen, 1997;Stepan, 1997).Unfortunately, studies of political elites have not gured prominently in studies of federalism. The complex relationship between elite structure and behavior on the one side and the effectiveness of federal institutions on the other has been little investigated. While a number of classic elite studies discuss politics in federal countries (e.g. Mills, 1956;Porter, 1965;Presthus, 1973;Higley et al., 1979;Hoffmann-Lange, 1991;Domhoff, 1998), they are not concerned with federalism per se. Except perhaps when elite recruitment is examined, studies of political elites in federal systems treat federalism as little more than a footnote.The many scholars of federalism, for their part, view its effectiveness as resulting from the 'right' institutional con guration, rather than from how political elites function. Either these scholars neglect the role of elites completely or they assign elites only implicit importance (e.g. ). Even comparativists who recognize that institutional performance depends heavily on elite functioning treat the elite variable as a kind of black box (Higley and Burton, 2000).This paper seeks to narrow the analytical gap between federalism and political elites. We assume that the causal arrow in federal systems generally runs from elites to institutions. We hold, in other words, that a well-functioning federal polity depends upon having the 'right' type of political elite con guration. We are aware that in such complex matters causal arrows often run in both directions, so we are also interested in understanding when and how federal institutions affect elite structure and functioning. First we canvas the