The strategic elites model of turnout argues that elites mobilize more when the probability of their effort deciding the electoral outcome is greater. Although the literature assumes that this probability depends solely on how close the election is, logically it depends jointly on how many votes are needed to affect the outcome (closeness) and on how many additional votes elite efforts are likely to garner (vote yield). Because the vote yield of mobilizational effort varies with the social capital of the district that elites face, the level of elite mobilizational effort (hence turnout) should depend interactively on closeness and social capital. The authors test their predictions using data from Japanese lower house elections for the years 1967-90. Japan is an interesting test case both because its (former) electoral system differs from that for which the model was first developed and because the literature clearly stresses the role of elite mobilization through social networks but does not examine the particular hypotheses advanced here.