Prior research has produced inconclusive results concerning the effects of negative campaigning. Researchers’ reliance on encyclopedic, even-handed measures of the tone of campaigns may help account for this inconsistency, for such measures are unlikely to reflect the way that most citizens process information about campaigns. Testing this argument by analyzing data from three 1998 gubernatorial campaigns, we observe a lack of consensus in citizens’ perceptions of these campaigns, a lack of convergence between citizens’ perceptions and social science-style classifications of these campaigns, and an array of biasing factors in citizens’ perceptions.