Attribute-framing bias reflects people's tendency to evaluate positively framed objects more favorably than the same objects framed negatively. Most theoretical accounts of this bias emphasized the role of positive- and negative-framing valence in the message, disregarding the quantitative information that typically accompanies it. To examine the role of both framing valence and detailed quantitative information in attribute-framing bias, we applied the distinction between gist and verbatim representations, as proposed by fuzzy-trace theory. We hypothesized that gist representations retain the framing valence used in the scenario, consequently eliciting biased positive or negative evaluations, whereas verbatim representations retain detailed quantitative information that allows for fine-tuning of the evaluations reflective of the magnitude of the target attribute. In 2 experiments, we compared precise presentations of different magnitudes using percentages and pie charts with vague presentations using verbal descriptions. A substantial attribute-framing bias was found for both the precise and vague presentation conditions, consistent with the hypothesis that the framing bias is driven by coarse and imprecise gist representation. Critically, however, the findings reveal higher correlations between evaluations and the magnitude of the target attribute in the precise presentation conditions (percentages and pie charts) compared with the vague verbal presentation. This finding suggests a process of fine-tuning of the evaluations when a detailed verbatim representation of the quantitative information is available. We discuss the findings in view of the distinction between gist and verbatim representations and propose future research to examine similar cognitive mechanisms accounting for biases in judgment and decision making.