This paper uses a sample of recent Senate election results and estimates vote equations that show challenger spending hurts, and incumbent spending helps, incumbent re-election. While both types of spending have diminishing returns, the effects are asymmetrical. Challenger spending is more productive at lower levels of spending, but incumbents can spend greater amounts more profitably than can challengers. These results can explain why Senate incumbents spend money, why they typically outspend their challenger, and why incumbents who can outspend their challenger would tend to be against spending limits or public financing. However, the results do not explain why incumbent spending does not "work" in House election equations. Jacobson and others have run countless linear and quadratic specifications that persistently show perverse effects for incumbent spending. These results are not affected by the procedural problem of logging observations that have a value of zero, and pose a genuine puzzle. There are other empirical results suggesting the idea that there are basic differences in the nature of elections between the House and Senate. For example, Grier and Carlson (1988) find that state-level economic conditions have a strong effect on individual Senate elections, while Owens and Olson (1980) find that district-level economic conditions have no effect on House elections. Since I show that there are a significant number of elections where incumbent spending does matter, and that simultaneity bias may not be a tenable explanation for results where incumbent expenditures do not matter, it may be time to take a new look at the House data or to develop a testable theory that can explain persistant empirical differences in the determinants of elections in the House and Senate.