What impact did the emergence of public opinion polls have on politicians and newspaper journalism? Based on interviews in the 1980s with American journalists and congressmen from the 1930s and 1940s, Susan Herbst argued that while "traditional" methods of assessing public opinion remained ubiquitous, attempts to quantify public opinion were also widespread. This paper offers a critique of her methods, questions the notion of a "transition", and reports a different set of findings. Based on an exhaustive examination of the Australian metropolitan press during the 1951 referendum, it shows the limited impact of polls on political journalism-indeed, on political calculus more generally-and how heavily Australian journalists and politicians relied on "rational" measures of long standing not discussed by Herbst-results of earlier referenda and of previous elections, as well as the betting odds. It also shows the importance of a range of "traditional" sources-the reception accorded party leaders at campaign rallies, politicians and campaign organisers' reports, the extent to which the referendum had divided parties, judgments about the popularity of state governments, and the activities of key interest groups-some of which, again, are not noted by Herbst.