Recent findings show that preschool children are selective with respect to whom they ask for information and whose claims they endorse. In particular, they monitor an informant's record of past accuracy or inaccuracy and use that record to gauge future trustworthiness. We ask if preschoolers also monitor the non-verbal cues of assent or dissent that bystanders display toward an informant's claims and use that information to gauge an informant's trustworthiness. In familiarization trials, 4-year-olds watched as two adult informants made conflicting claims regarding the name of an unfamiliar object. Two adult bystanders consistently signaled assent – via nods and smiles – to the claims of one informant, and dissent – via head shakes and frowns – from the claims of the other informant. When invited to endorse one of the two claims, 4-year-olds mostly agreed with the informant who had received bystander assent. Thus, in the absence of background knowledge about an object's name, children use third-party non-verbal signals to assess the accuracy of conflicting labels. On subsequent test trials, the informants again made conflicting claims about novel object names, but in the absence of the two bystanders. Despite the lack of any informative bystander signals, children with more advanced understanding of mental states continued to display greater trust in the informant who had received bystander assent in the earlier trials.