HE original formulation of the two-step flow hypothesis by Lazarsfeld et al. (1948) may have been the most influential and durable product of their classic study. Not only did the formulation redirect the field of mass communications research; its emphasis on the power of primary groups, and informal, interpersonal, and gemeinschaft-\ike ties introduced an empirically based and powerful alternative to theories of mass society which viewed an atomized population as politically manipulated by the mass media (DeFleur, 1970;Kornhauser, 1953).This article examines the two-step flow hypothesis in the context of voting behavior in a recent national election, the behavior the original hypothesis attempted to explain. It attempts to review the several criticisms of this statement of the relation between the mass media and social networks that was formed in an era before more advanced political media systems (particularly television) became available and before a more educated and media-dependent public had emerged. 1 1 For example, Lazarsfeld et al. (1948) reported a greater prevalence of interpersonal discussions than mass media usage during a typical day in the 1940 presidential campaign. In a survey conducted during the last weeks of the 1972 campaign, Richard Hofstetter (personal communication) found 18 percent of a national sample of over 1,000 adults had talked with Abstract Research subsequent to the classic statement of a two-step flow-of-influence hypothesis has isolated a number of inaccuracies in the hypothesis as originally formulated. Analysis of a question series on interpersonal influence, included in the 1968 national election study of the Center for Political Studies, reveals further difficulties with the hypothesis. These analyses further suggest the existence of two distinct patterns of mass media and interpersonal influence at work in separate segments of the electorate. One process operates solely through interpersonal channels, in which the mass media's role is unclear. The second proceeds directly through exposure to the mass media, particularly newspapers. Some ways in which the study results are consistent with the original hypothesis are discussed.John P. Robinson is Professor of Communication and Director of the