It is widely acknowledged that political opinions are not simply positive or negative, but are often simultaneously positive and negative, or "ambivalent." Although there is evidence that ambivalence influences the dynamics of public opinion on policy issues, little is known about its role in contributing to electoral decision making. Using National Election Studies data from 1980 to 1996, I examine the consequences of ambivalence toward presidential candidates for electoral judgment and choice. Results revealed that ambivalence created instability in candidate evaluations, substantially delayed the formation of citizens' voting intentions, conditioned the influence of both personality assessments and issue proximity on summary candidate evaluation, and ultimately weakened the prediction of vote choice. Throughout the analyses, the effects of ambivalence were independent of and typically larger than those of partisanship strength, information, education, and attitude strength, and could not be meaningfully accounted for by any of these factors. In broad terms, ambivalence would appear to capture a unique and fundamental -although to date largely ignored -aspect of mass belief systems and electoral choice.