This article draws upon research on negativity bias and motivated reasoning to develop a theory explaining how and why partisan bias moderates the relationship between impressions of character weakness and evaluations of presidential candidates. Impressions of character weakness exist at the aggregate level, which center on the perceptions of the public as a whole, and the idiographic level, which center on individual differences in perceptions. The theory predicts the more strongly people identify with the opposition party of a presidential candidate, the stronger the relationship between impressions of character weakness and candidate evaluations. Analysis of 1984-1996 National Election Study data provides partial support for the theory. The results show partisan opponents rely more on perceptions of character weakness than partisan supporters when evaluating presidential incumbents and salient nonincumbents in the aggregate level models and that partisanship does not promote reliance on perceptions of character weakness in the idiographic level models.