In three studies we investigated the role of bottom-up information processing in attitudinal judgment. Overall, the results confirm our expectations and show that people are faster in judging attributes underlying their attitude towards the object than in generating or 'computing' their overall attitudinal response. As predicted, respondents who selected more attributes as important to their attitude needed more time to integrate these attributes in order to come to an overall attitudinal response. Moreover, ambivalence was also related to decreased response times of the overall attitudinal response. We argue that the main reason for this is that non-ambivalent attitudes are generally based on evaluatively congruent attributes, while ambivalent attitude-holders need to integrate evaluatively incongruent at- tributes into an overall judgment. Implications for research on attitude structure and ambivalence are briefly discussed. A simple question like: 'Would you like a slice of blueberry pie?' may evoke a simple answer such as 'Great!'. Other people may want to think twice before accepting the offer. The process through which people arrive at such an overall evaluative judgment may thus vary considerably. Some provide an immediate answer. Others might engage in more elaborate thinking and consider a number of aspects specific to blueberry pie, such as its taste, number of calories etc., before reaching a decision. We focus on these processes and address the issue of how people come to overall evaluative judgments. A number of theories emphasize the role of deliberate, controlled information processing, and generally assume that evaluative judgements are based on judgments of more specific attributes. These judgments are expected to be integrated into an overall attitudinal response. Examples are expectancy-value (EV) models such as those of Fishbein and Ajzen (Ajzen, 1991; Fishbein and Ajzen, 1975), and Anderson's information integration theory (Anderson, 1971). Although these theories have been