It is commonly argued in the media that a presidential candidate will be helped in a state by having a governor of the same party in office. However, there is little research to support this claim. To address this question we use a regression discontinuity design. The basic idea behind this is that in very close elections the party of the governor is decided essentially by a coin flip. Focusing on these very close elections therefore allows us to estimate the causal effect of gubernatorial party control. We show that a presidential candidate is not helped, but in fact hurt, by having a governor from the same party. On average, winning the governor’s election leads to a 2–3 percentage point reduction in a state’s presidential vote share in the following election. Using a similar methodology, we also show that voters punish the presidential party when voting for governor in midterm years. Having established these relationships, we explore why this is the case. One possible explanation is a variation of the ideological balancing argument, whereby voters’ choices for one office are conditional on which party holds office at a different level.