We examined how the format in which uncertainty information is presented affects two biases in humans' choice behavior. In a computer task, participants were given four common-ratio effect and four common-consequence effect problems in each of four different formats. In these problems, uncertainty information was described, as percentages (e.g., 80%) or as frequencies (e.g., 16/20), or was experienced, either serially (20 outcomes shown one at a time) or simultaneously (20 outcomes all shown at once). Presenting information as percentages attenuated the common-ratio effect and augmented the common-consequence effect, which suggests that these biases have different underlying mechanisms. Participants' percentage estimates of outcome likelihoods did not differ according to the format in which the information was presented; however, participants' nonverbal estimates of outcome likelihoods differed across formats. The results suggest that uncertainty information presented as percentages is processed differently than the same uncertainty information presented in other formats.