Many scholars address the indirect effects presidents have on midterm election results by examining the “midterm loss” phenomenon, presidential coattails, negative voting, and the “referendum” thesis. However, very little research investigates the direct effect that presidential campaigning has on congressional candidates prospects for victory. This study adds to this growing literature by exploring presidential campaigning in the 2002 and 2006 U.S. Senate midterm elections. Our investigation makes two important contributions to previous research. First, we explicitly model the strategic decisions presidents make in visiting states in order to get a better estimate of presidential impact that accounts for selection bias. Second, we take advantage of a natural experiment between 2002 and 2006 to test the importance of presidential popularity in driving campaign effects. Although we expect presidents to primarily campaign in states where he believes he will be effective–an efficiency argument–we argue that the size of that effect depends very clearly on his own popularity. We discuss the implications of these results for understanding the role of the president in modern American electoral politics.