A landmark finding in recent research on electoral behaviour is that voters anticipate the postelection bargaining process among potential members of the governing coalition, and that these anticipated policy agreements inform their vote choice. In this article, this finding is qualified by arguing, and then showing empirically, that when the expected policy change after the elections is marginal or non-existent, ceteris paribus, ‘simple’ proximity voting should prevail. The argument is tested by using two different but complementary research strategies applied to an individual-level data set constructed from electoral surveys in 28 countries over a 20-year period, and two recent national surveys in which respondents were directly asked to predict the potential coalition government after the elections. Both strategies provide support for the hypothesis and have important implications for the understanding of the consequences of government alternation on voting behaviour and political representation more broadly.