Studies have shown that participants can adequately take into account several cues regarding the weight they should grant majority opinions, such as the absolute and relative size of the majority. However, participants do not seem to consistently take into account cues about whether the members of the majority have formed their opinions independently of each other. Using an evolutionary framework, we suggest that these conflicting results can be explained by distinguishing evolutionarily valid cues (i.e. they were present and reliable during human evolution) from other cues. We use this framework to derive and test five hypotheses (H1 to H5). Our first three experiments reveal that participants discount majority opinion when the members of the majority owe their opinions to the same hearsay (H1), owe their opinions to having perceived the same event (H2), or owe their opinions to a common motivation (H3). Experiment 4 suggests that, by contrast, participants do not discount majority opinion when the members of the majority owe their opinions to sharing similar cognitive traits (H4). Finally, Experiment 5 suggests that participants adequately discount majority opinion when one of the members of the majority is untrustworthy (H5). This set of experiments shows that participants can be quite skilled at dealing with informational dependency, and that an evolutionary framework helps make sense of their strengths and weaknesses in this domain.