Given President Bush's popularity among relatively poor rural residents and lack thereof among wealthier urban dwellers in the 2004 presidential election, analysts have suggested that voters contradicted their economic self-interests. We investigate whether this conventional wisdom implied an absence of economic voting. Using exit poll data, we estimate whether a change in previous four-year financial status affected the propensity to vote for Bush. The main econometric concern is that underlying preferences for Bush might dictate financial status change responses. Beyond income and several other demographic variables, therefore, the regressions hold constant indicators for state and congressional district, religious affiliation, political philosophy and party, and Iraq war support. Even further controlling for approval of Bush's job performance, economic voting is statistically and quantitatively significant. Effects are asymmetric, with status worsening hurting Bush more than status improvement helped, and persist even among subgroups that provided particularly strong or weak support for Bush.