Although presidential campaigns have been mythologized in literature and cinema, most theories of elections relegate them to a secondary role, presuming they have little effect on outcomes. Direct tests of campaigning's influence on electoral college votes are rare, mostly because statewide data on the allocation of resources and voters' preferences have been hard to obtain. Many studies suggest a minimal effect, but it is possible that a more significant influence might be found with better data on the key dependent and independent variables. This study uses data on presidential candidates' appearances and television advertising purchases to conduct cross-sectional and pooled time-series analyses of their influence on statewide outcomes in 1988, 1992, and 1996. The data demonstrate that, despite the conditioning influence of other factors, campaigning affected statewide preferences as well as the electoral college vote.