In this article, the reason why incumbent spending is less effective than challenger spending is explained. The argument is that incumbent spending efficiency depends on the marginality of seats: safe incumbent spending is less effective than marginal incumbent spending, since safe incumbents have to buy fewer extreme voters, whereas marginal incumbents can easily buy a larger number of swing voters. The analysis of the US Senate elections between 1974 and 2000 shows that safe incumbent spending is less effective than challenger spending, but marginal incumbent spending is not. The analysis also shows that the previous finding of less effective incumbent spending is largely due to the fact that the data for marginal and safe incumbents have been aggregated.